Paving the Great Way: Vasubandhu’s Unifying Buddhist Philosophy

2. AKBh IV.3 (194.14–18): nāsti saṃsthānaṃ dravyata iti sautrāntikāḥ | ekadiṅmukhe hi bhūyasi varṇa utpanne dīrghaṃ rūpam iti prajñapyate | tam evāpekṣyālpīyasi hrasvam iti | caturdiśaṃ bhūyasi caturasram iti | sarvatra same vṛttam iti | evaṃ sarvam | tadyathā ’lātam ekasyāṃ diśi deśāntareṣv anantareṣu nirantaramāśu dṛśyamānaṃ dīrgham iti pratīyate sarvato dṛśyamānaṃ maṇḍalam iti | na tu khalu jātyantaram asti saṃsthānam.

3. As Anacker (1998:130–131, n12) has noted, lham pa is probably a mistaken translation of vṛtta, which means “circle,” here, as it is in two passages in the Tibetan translation of the AKBh (Hirakawa et al. 1973:340).

4. Anacker (1998:131, n12) takes these as “concave” and “convex.”

5. KSP D.135r7–v4: phyogs gcig gi sgor ’dus pa mang por snang ba la ni ring po’i blo ’byung | thung du la thung [135v] du’i blo ’byung | thams cad nas mnyam bar snang ba la ni lham pa’i blo ’byung | khor yug nas mnyam pa la ni zlum po’i blo ’byung | dbus na mang po la ni mthon po’i blo ’byung | nyung ba la ni dma’ ba’i blo ’byung | phyogs gcig gi skor snang ba la ni phya le ba’i blo ’byung | phyogs sna tshogs kyi sgo la ni phya le ma yin pa’i blo ’byung ngo |. … kha dog kho na yul gi khyad par gnas pa la ring po la sogs pa’i blo dag ’byung ste | dper na shing dang bya dang | grog ma la sogs pa’i dngar ka dag la ’byung ba lta bu ste.

6. This may not be true for all theories of language, but it ought to be for the AKBh. AKBh I.14 (10.16) has the third skandha, saṃjñā, “idea,” as the grasper of concepts, including such examples as “long” (dīrgha) and “short” (hrasva).

7. AKBh II.47a–b (80.24–25): vāṅnāmni pravarttate nāmārthaṃ dyotayati naiva ghoṣamātraṃ vāgyena tu ghoṣeṇārthaḥ pratīyate sa ghoṣo vāk.

8. AKBh II.47a–b (81.21): atrārtheṣu kṛtābadhiḥ śabdo nāma …

9. For this reason, Vasubandhu’s position in the AKBh might be subject to Arnold’s (2012:140–141) critique of Dharmakīrti that while he would like to have reduced linguistic events to causal stories, in fact his causal stories always appeal at some point to a human subject. The KSP is not, however, subject to this critique.

10. By calling the false projection “subjectivity,” I am emphasizing the subjective component in the denial of duality here, but of course duality has two sides (which is the point of its name), and Vasubandhu does spend a good deal of his time disproving the mistaken construction of mental objects as well as subjects. The error of “duality,” though, is not the mistake of reifying any particular mental object or momentary subject, per se; it is, rather, a mistake implicit in the structure of awareness and language. Modern theorists following Brentano talk about the “intentionality” of language and consciousness, by which is meant the “aboutness” that makes consciousness and meaning what they are; consciousness is always consciousness of something, and words are always about something. For Vasubandhu’s Yogācāra, as we will see, the term “duality” (dvaya) picks out this “aboutness” of consciousness and language and calls it a fundamental error that appears in every conceptual-linguistic moment until enlightenment. Things appear “dual” when in fact they are “only cause and effect.”

11. See Schmithausen 1984, Aramaki 2000, and Buescher 2008 on this passage. It is Schmithausen’s thesis that this was the “initiatory passage” of the vijñapti-mātra, a view with which Buescher disagrees. See chapter 3, note 71. As will become clear, the key point for me here is that the term vijñapti-mātra is being used to deny the distinction between mind and mental objects.

12. Here I am referring to Vasubandhu’s citation of the SNS at the pinnacle of his scriptural hermeneutics (see chapter 4). For his use, and transformation, of the discussion of the elephant analogy from SNS I, see Gold 2006.

13. SNS VIII.7 (91.2–3): ci’i phyir tha dad pa ma yin zhe na | gzugs brnyan de rnam par rig pa tsam du zad pa’i phyir te.

14. SNS VIII.7 (91.7–8): sems de nyid kyis sems de nyid la ji ltar rtog par bgyid lags.

15. SNS VIII.7 (91.8–17): byams pa de la chos gang yang chos gang la ’ang rtog par mi byed mod kyi | ’on kyang de ltar skyes pa’i sems gang yin pa de ni de ltar snang ngo | byams pa ’di lta ste dper na | gzugs la brten nas me long gi dkyil ’khor shin tu yongs su dag pa la gzugs nyid mthong yang gzugs brnyan mthong ngo snyam du sems te | de la gzugs de dang | gzugs brnyan snang ba de don tha dad par snang ngo || de bzhin du de ltar skyes pa’i sems de dang | ting nge ’dzin gyi spyod yul gzugs brnyan zhes bya ba gang yin pa de ’ang de las don gzhan yin pa lta bur snang ngo.

16. AKBh 31.11–13: kim idam ākāśaṃ khādyate | cakṣur hi pratītya rūpāṇi cotpadyate cakṣurvijñānam | tatra kaḥ paśyati ko vā dṛśyate | nirvyāpāraṃ hīdaṃ dharmamātraṃ hetuphalamātraṃ ca.

17. Dignāga claims that “self-cognition” (svasaṃvitti) must be present within a mental event if it is to account for the subsequent arising of memory. If by this he means self-cognition to be a separate entity, he would seem to have forgotten, or betrayed, this argument of Vasubandhu’s. The doctrine of self-cognition is elaborated upon by Dharmakīrti and his successors, and is a topic of wide-ranging discussion in the contemporary study of Buddhist philosophy. See, for instance, Arnold 2012: esp. 158–198, Coseru 2012, Williams 1998b, and Yao 2005. Vasubandhu does not take up the problem introduced and expanded upon by these authors, traditional and modern. In addition to his evident displeasure with the argument from memory, see his response to the issue of the Buddha’s omniscience, at AKBh VII.18, which I discussed in chapter 3, note 42. Vasubandhu’s view is straightforwardly that subjectivity is a false construction, which falls away at enlightenment. Once the duality between subject and object dissolves, the self-awareness of consciousness does not arise, since nondual awareness is inconceivable. Memory, however, is sufficiently supported by ordinary causality. The prior awareness is a support (āśraya) for the later consciousness, just like any other mental consciousness. See below for further discussion of Vasubandhu’s disproof of subjectivity, and for his treatment of the memory argument in the Viṃś.

18. On this move from the meditative context to a global metaphysical claim, see Griffiths 1991 and Schmithausen 2005.

19. SNS VII.8 (91.18–24): bcom ldan ’das sems can rnams kyi gzugs la sogs par snang ba sems kyi gzugs brnyan rang bzhin du gnas pa gang lags pa de ’ang sems de dang tha dad pa ma lags zhes bgyi ’am | bka’ stsal pa | byams pa tha dad pa ma yin zhes bya ste | byis pa phyin ci log gi blo can rnams ni gzugs brnyan de dag la rnam par rig pa tsam de nyid yang dag pa ji lta ba bzhin mi shes pas phyin ci log tu sems so.

20. Viṃś 3.2: mahāyāne traidhātukaṃ vijñaptimātraṃ vyavasthāpyate. See Rahder 1926:49 for the Daśabhūmikasūtra source.

21. In addition to sources mentioned in the “Doctrinal Positions and Works” section of Gold 2011, see Siderits 2007:146–173 for an insightful summary of the argument discussed in this section, and Kapstein 2001b for a close analysis of the argument about perceptual objects. Beginning here, several pages from Gold 2011 have been edited for integration into the present work, and are printed here with the kind permission of the Stanford Encyclopedia of Philosophy.

22. See Gold 2011 for a discussion of the question of Yogācāra as “idealism.”

23. The work is generally described as a defense of the Yogācāra school of Mahāyāna Buddhism, in contradistinction to the Madhyamaka school. But during Vasubandhu’s time there was no division between Yogācāra and Madhyamaka; on the contrary, the SNS claims to “clarify the intent” of the Mahāyāna overall, and eliminate disputation. Although Asaṅga and Vasubandhu do call into question certain alternative interpretations of Mahāyāna (see Gold forthcoming), these intra-Mahāyāna disputes do not arise in the Viṃś. My understanding, therefore, is that the Viṃś sees the defense of “appearance only” as a clarifying defense of Mahāyāna overall.

24. Failure to appreciate this fact lies at the root of many of the most common critiques of Vasubandhu’s argument, both traditional and modern. See for instance Sarachchandra 1976 and Feldman 2005.

25. See Gold 2006, and sources cited therein, for further discussion of Vasubandhu’s use of the dream example.

26. This is what Siderits (2007:157) calls Vasubandhu’s use of the “Principle of Lightness,” or Occam’s razor. It may also be recast as a declaration that the doctrine of appearance-only solves the continuity problems associated with the causal relationship between physical and mental entities.

27. The alternative to this quandary would be to deny karma and be stuck with a different quandary, what Owen Flanagan (2009) calls “the really hard problem” of how to find meaning in a physical world whose causes are not indexed to human intentions.

28. Kapstein (2003b) introduces and translates Vasubandhu’s mereological argument in full. I provide a bit more detail in Gold 2011 than I do here.

29. Viṃś 6.29–30: na tāvad ekaṃ viṣayo bhavaty avayavebhyo ’nyasyāvayavirūpasya kvacid apy agrahaṇāt.

30. Viṃś 6.30–7.1: nāpyanekaṃ paramāṇūnā pratyekam agrahaṇāt.

31. Viṃś 8.22–9.1: pramāṇavaśād astitvaṃ nāstitvaṃ vā nirdhāryate sarveśāṃ ca pramāṇānāṃ pratyakṣaṃ pramāṇaṃ gariśṭam ity asaty arthe katham iyaṃ buddhir bhavati pratyakṣam iti | pratyakṣabuddhiḥ svapnādau yathā [16a–b1] vināpy artheneti pūrvam eva jñāpitaṃ | sā ca yadā tadā | na so ’rtho dṛśyate tasya pratyakṣatvaṃ kathaṃ mataṃ || (16) [16b2–d] yadā ca sā pratyakśabuddhir bhavatīdaṃ me pratyakṣam iti tadā na so ’rtho dṛśyate manovijñānenaiva paricchedāc cakṣurvijñānasya ca tadā niruddhatvād iti | kathaṃ tasya pratyakṣatvam iṣṭaṃ | viśeṣeṇa tu kṣaṇikasya viṣayasya tadānīṃ niruddham eva tad rūpaṃ rasādikaṃ vā.

32. Viṃś 16–17, 9.1–8: nānanubhūtaṃ manovijñānena smaryata ity avaśyam arthānubhavena bhavitavyaṃ tac ca darśanam ity evaṃ tad viṣayasya rūpādeḥ pratyakṣatvaṃ mataṃ | asiddham idam anubhūtasyārthasya smaraṇaṃ bhavatīti | yasmāt | uktaṃ yathā tadābhāsā vijñaptiḥ [17a–b1] vināpy arthena yathārthābhāsā cakṣurvijñānādikā vijñaptir utpadyate tathoktam | smaraṇaṃ tataḥ | [17b2] tato hi vijñapteḥ smṛtisaṃprayuktā tatpratibhāsaiva rūpādivikalpikā manovijñaptir utpadyata iti na smṛtyutpādād arthānubhavaḥ sidhyati.

33. As discussed in chapter 4, Vasubandhu preserves the notion of perception as a valid means of knowledge, especially for a Buddha, but in several places—as here—he considers it essential to use reason to test and contextualize what we think we perceive.

34. Vasubandhu does not use the term “emptiness” in this characteristic, Mahāyāna inflection in his corpus, except in the MAVBh (on which, see below), so it should not be assumed that this now common usage would be natural to him. But there is a clear continuity between my usage here, and below, of the term “emptiness,” based in the MAV, and the lengthier but conceptually equivalent term “lack of own-being” (ngo bo nyid ma mchis pa) which appears in the SNS—a text that Vasubandhu cites and uses extensively. I have taken the liberty of simply referring to the concept in this shorter, more familiar, way. For discussion of this continuity, see below and Gold (forthcoming).

35. This is not to say that Vasubandhu was skeptical of Buddhist doctrine overall—quite the contrary—but he did believe that even Buddhist doctrine needed to be tested, to see how, and when, it was most literally true, and when it needed to be taken with a grain of salt. See chapter 4.

36. Sthiramati’s famous commentary on the Thirty Verses does include extensive philosophical defenses of the doctrines advanced in the verses. But Vasubandhu’s own arguments, if he had any, are lost. The verses include only definitions and lists of terms. The Three Natures Exposition is also a systematic survey with none of Vasubandhu’s (or anyone else’s) commentarial defenses. It is possible, of course, that one or both of these works is/are misattributed to Vasubandhu. The Three Natures Exposition is certainly quite different in style from his other works. Philosophically, however, it follows quite logically from his other works, as I argue here.

37. I have not made Vasubandhu’s approach to the path a focus in this work, though it is surely a scholarly desideratum. Tzohar 2011 contains an insightful discussion of the Triṃś, which parallels my analysis here.

38. TSN 154.1–6: kalpitaḥ paratantraś ca pariniṣpanna eva ca | trayaḥ svabhāvā dhīraṇāṃ gaṃbhīraṃ jñeyam iṣyate || (1) yat khyāti paratantro ‘sau yathā khyāti sa kalpitaḥ | pratyayādhīnavṛāt kalpanāmātrabhāvataḥ || (2) tasya khyātur yathākhyānaṃ yā sadāvidyamānatā | jñeyaḥ sa pariniṣpannaḥ svabhāvo ‘nyathātvataḥ || (3).

39. The interconnected, unified character of the three natures is especially clear in verses 10–16, where the identity of each nature is played against the others in such a way as to display their conceptual dependence. See below, and the notes to appendix G, where I discuss these interconnections in detail.

40. TSN 154.7–8: tatra kiṃ khyāty asatkalpaḥ kathaṃ khyāti dvayātmanā | tasya kā nāstitā tena yā tatrādvayadharmatā || (4).

41. Although I am satisfied with my past translation of the term dvayātmanā (here “as dual”) with the phrase “as a dual self,” this was something of a creative translation, and here I do not want to mislead the reader as to my intentions, which I argue for in detail in Gold 2006. The word ātman means “self”; it is what is denied by the Buddha. Thus the word dvayātmanā means, literally, “with a dual self” or “with the self of duality.” But in addition to the meaning “self,” the term ātman also means, more broadly, “nature” or “essence,” so it is quite often used idiomatically to mean “with the nature of x.” This is a particularly common meaning in śāstric commentarial literature, but it may appear in verse as well. For this reason, most translators of the TSN read the verse to say that the unreal fabrication appears “as dual,” or “with a dual nature,” leaving aside any implicit meaning of “self.” In the next line, however, the verse uses advayadharmatā, which I translate here as “there being no duality,” to refer to the appearance’s fundamental lack of a false, dual nature. The alternation between ātmanā (lit. “with a nature”/“with a self”) and dharmatā (“true nature”) enlivens the dead metaphor of ātmanā. There is an undeniable pun on the term ātmanā. The “dual self” (dvayātman) is the essential nature of the duality between subject and object, which generates, for living beings, the false belief in a self (satkāyadṛṣṭi—see chapter 3). It is directly contrasted with the “nondual true nature” (advayadharmatā). This punning meaning on “self” makes sense of both instances where Vasubandhu uses the term ātmanā in the TSN, as opposed to the instances where instead he uses the term ātmika, which has the same idiomatic meaning, but no implicit pun. As a further defense of this reading, I would point to the following passage from AKBh II.46, where Vasubandhu uses the term ātmanā with exactly the same implicit pun—the falsely imposed “nature” is the nature of a self: “How, when the dharmas of birth of birth, etc., have been made unified, does a dharma come about as a ninth nature (ātmanā)?” (AKBh 75.15–16: jātijātyādīnāṃ caikatra dharme katham kṛtā ātmanā navamo hi dharma utpadyate.) Since this is only a pun, though, and not a direct philosophical statement, here I leave the “dual self” in the notes.

43. As I will discuss below, this amounts to saying that in order to free oneself of the false construction of self, to free oneself of the false imposition of “duality,” one must overcome the experience of consciousness as “intentional” in the philosophical sense of being “about” its experiential objects.

44. See Gold 2006 for detailed treatment of duality in early Yogācāra. See sources cited therein for a diversity of opinions on this complex topic.

45. AKBh 31.15–16. Discussed in chapter 3.

46. DhDhVV 71–72: chos nyid kyi mtshan nyid ni | gzung ba dang | dzin pa dang | brjod par bya ba dang | rjos par byed pa khyad par med pa’i de bzhin nyid de ni chos nyid kyi mtshan nyid do | gzung ba dang ’dzin pa’i khyad par med pa gang yin pa dang | brjod par bya ba dang [72] | rjod par byed pa khyad par med pa gang yin pa de ni de bzhin nyid yin la | de yang chos nyid kyi mtshan nyid yin no || khyad par med pa nyid ni rim pa ji lta ba bzhin gnyis su med pa’i phyir dang | brjod du med pa’i phyir te | gang la khyad par yod pa ma yin pa de ni khyad par med pa zhes bya bas so.

47. Although I have presented conceptual duality as distinct from conceptual-linguistic duality, it seems possible to me that conceptual-linguistic duality is always the target for Yogācāra thinkers, and that the ostensible bifurcation of the world into x and not-x is considered a “duality” only because it is always the case that one or the other is intended. Since Vasubandhu’s distinctive innovation is to foreground perceptual duality, I will not pursue the evidence of conceptual vs. conceptual-linguistic dualities here. Let me also acknowledge that the root verse in this passage does equate the nondifference of “grasper and grasped” with the nondifference of “expression and expressed,” and so does seem to be a central inspiration for Vasubandhu’s commentarial uniting of the perceptual with the conceptual-linguistic forms of nonduality, as I will discuss below.

48. TSN 154.9–10: asatkalpo ‘tra kaṣ cittaṃ yatas tat kalpyate yathā | yatā ca kalpyaty arthaṃ tathātyantaṃ na vidyate || (5).

49. TSN 154.11–18: tad dhetuphalabhāvena cittaṃ dvividham iṣyate | yad ālayākhyavijñānaṃ pravṛttyākhyaṃ ca saptadhā || (6) saṃkleśavāsanābījaiś citatvāc cittam ucyate | cittam ādyam dvitīyaṃ tu citrākārapravṛttitaḥ || (7) samāsato ‘bhūtakalpaḥ sa caiṣa trividho mataḥ | vaipākikas tathā naimittiko ‘nyaḥ prātibhāsikaḥ || (8) prathamo mūlavijñānaṃ tad vipākātmakaṃ yataḥ | anyaḥ pravṛttivijñānaṃ dṛśyadṛgvittivṛttitaḥ || (9).

50. In this way, the storehouse consciousness may be read to provide a far more robust account of personal continuity than could be achieved by simply calling the self an illusion. Conventionally speaking, we are who we are from moment to moment, and from life to life, because of the storing and subsequent activation of this subliminal store consciousness. Some Yogācāras even say that the purificatory transformation of this storehouse consciousness into a series of seeds devoid of negativities constitutes the attainment of Buddhahood. So the storehouse consciousness certainly accounts for some aspects of personal identity over time. Since the karmic “seeds” are momentary, like everything else, and the ālayavijñāna is nothing but those seeds, there is no ultimately real identity to the storehouse. But it provides a useful analogy.

51. TSN 156.3–10: māyākṛṃ mantravaśāt khyāti hastyātmanā yathā | ākāramātraṃ tatrāsti hastī nāsti tu sarvathā || (27) svabhāvaḥ kalpito hastī paratantras tadākṛtiḥ | yas tatra hastyabhāvo ‘sau pariniṣpanna iṣyate || (28) asatkalpas tathā khyāti mūlacittād dvayātmanā | dvayam atyantato nāsti tatrāsty ākṛtimātrakam || (29) mantravan mūlavijñānam kāṣṭhavat tathatā matā | hastyākāravad eṣṭavyo vikalpo hastivad dvayam || (30).

52. Kaplan (1990) provides a useful modern analogy, the hologram, which was of course not available to Vasubandhu. In Gold 2006 I argue that Vasubandhu chose the magical illusion analogy for its ability to express the illusion of selfhood.

54. MAVBh 17.4: vaktāraṃ cāsmadādibho. D’Amato (2012) provides an excellent translation and study of the MAV/MAVBh. I have benefited from his readings, but I have added changes of my own.

55. This is not to say that Vasubandhu was among the first to hear the text. The commentary allows the interpretation that its author was one among several people to whom Asaṅga spoke the text, successively, at different times.

56. I adopt a rather simplistic naïveté for the sake of clarity. The author of the MAV may indeed have been someone other than Asaṅga. Perhaps Asaṅga found the text in a cave, or forgot where he had originally heard it after remembering it in a dream—either of which might account for his faith in the revelation. Perhaps his real master swore him to secrecy.

57. MAV I.1 (17.16–17): abhūta-parikalpo ’sti dvayan tatra na vidyate | śūnyatā vidyate tv atra tasyām api sa vidyate || (I.1).

58. The term “emptiness” (śūnyatā) is the famous, central term of art for the Madhyamaka tradition founded by Nāgārjuna. It has been argued that the opening verses of the MAV are an attempt to provide a reading of this term from a Yogācāra perspective, but this is hardly an arbitrary shift in terminology; the MAV uses the term “emptiness” to refer to what the SNS (VII.31 [86.1]) calls “lack of own-being” (ngo bo nyid ma mchis pa).

59. MAVBh 18.1–4: tatrābhūtaprikalpo grāhya-grāhaka-vikalpaḥ | dvayaṁ grāhyaṁ grāhakañ ca | śūnyatā tasyābhūtaparikalpasya grāhya-grāhaka-bhāvena virahitatā | tasyām api sa vidyata ity abhūtaparikalpaḥ.

60. This expression is ordinarily translated “subject and object.” D’Amato (2012) does so, and I am not intending to register a complaint about that translation. I only preserve the more technical vocabulary so as to highlight the terms as a grammatical pair, as I explain below.

61. Admittedly, sometimes the term dvaya in the verses just refers to “two” of something (I.25, II.11). Yet there are several instances in which the verse text of the MAV uses dvayam to describe something metaphysically basic, and false. In those cases, we generally find it, as in this case, in close proximity to bhāva–abhāva or sat–asat; and these are exactly the contexts in which Vasubandhu glosses the term as “grasper and grasped” (I.1–2, I.13). In the TSN, we find the pattern of dvayam paired with sat–asat and bhāva–abhāva repeated over and over, but no use of grāhya–grāhaka. We do, however, find the use of dṛṣyadṛgvitti, which has the same meaning, as we will see, of consciousness and its object.

62. MAV I.2a–b (18.8): na śūnyaṁ nāpi cāśūnyaṁ tasmāt sarvvam vidhīyate.

63. A quick example on hand from the Vajracchedika-sūtra (Conze 2001:34–35): “Great, O Lord, great, O Well-Gone, would that Heap of merit be! And why? Because the Tathagata spoke of the ‘heap of merit’ as a non-heap. That is how the Tathagata speaks of ‘heap of merit.’ ”

64. In none of these cases is the pair of terms referred to as a dvayam, a “duality,” though they are referred to as “twofold” (dvidhā, at V.24)—they are, after all, two words. There is also an instructive use of the term grahaṇa at III.20.

65. This provides a parallel to TSN 9 if we read dṛśyadṛgvitti as three separate entities.

66. MAV III.17c–d (45.9): grāhaka-grāhya-tad-grāha-bījārthaś cāparo mataḥ.

67. MAV III.4–5a (38.10–12): samāropāpavādasya dharmma-pudgalayor iha | grāhya-grāhakayoś cāpi bhāvābhāve ca darśanaṁ || (III.4) yaj-jñānān na pravartteta tad dhi tatvasya lakṣaṇaṁ.

68. Each also appears in the lists from MAV V.23–26.

69. MAV III.16b1 (44.5): ātmadarśanam. D’Amato 2012:152.

70. At AKBh II.59, Vasubandhu argued against grahaṇa as a mode of causality at a temporal distance. This is a different kind of false conception of “grasping,” but, to characterize it as yet another version of this same illusion of false causality, it too expresses Vasubandhu’s preference for causality understood in direct, causal lines.

71. A final meaning for grahaṇa that may be worth considering as we attempt to make sense of the attribution of the grāhya–grāhaka gloss to Vasubandhu is the meaning given the term by the Vaibhāṣikas in their discussions of causality and temporality. Vasubandhu had critiqued a vagueness in Vaibhāṣika terminology that seemed to allow for the same entity to be present at two separate moments (see chapter 2). According to Akimoto (2004), the Vaibhāṣika Saṃghabhadra responded to Vasubandhu’s critique by insisting that entities exist in the past just until the moment of their “taking” (grahaṇa/pratigrahaṇa) their causal result, when they become present—thus removing the apparent problem. With such a view in mind, we might see Vasubandhu’s gloss on “nonduality” as grāhya–grāhaka as an updating of his AKBh argument, deploying a Yogācāra counter to the Vaibhāṣika terminological fix. For if causal series are by their nature “nondual,” there is no possibility that an entity could be a “taker” that “takes” its causal result in order to become present. Once present, it is a different entity than it was when past; the series of entities are not “taker and taken,” but merely cause and effect. This familiar argument would work quite well against the revised Vaibhāṣika view; but it is only a hypothetical. I have never seen it in any Buddhist text.

72. This claim is based upon a reading of deep, conceptual similarities and motives, but I do not hold that this counts as a “scientific proof” that the extant Yogācāra commentaries were necessarily composed by the same hand as the AKBh. My first chapter is intended to call such certainty into question. My point is, rather, that there is a distinctive, common complex of ideas and philosophical motivations identifiable across the works discussed here. As of yet I see no good reason to deny this “Vasubandhu complex” a single authorship, but should that become necessary, the conceptual nexus will remain. Buescher (2013) displays stylistic differences between MSABh and AKBh as evidence against a common authorship, but as I have said above, I do not consider this sufficient to rule out common authorship. As Corless (1989) argues, we are right, after all, to attribute both the Tractatus and the Philosophical Investigations to the same “Wittgenstein.”

73. On the contrary, although he defends the use of perception and inference in order to clarify our understanding, from the perspective of ultimate reality, he is skeptical of all “epistemic means” (pramāṇa). In the VyY he cites approvingly the scripture that says, “If the sensory organs were epistemic means, there would be nothing to do on the noble path.” See appendix E.

74. Pace Sellars (1997), who expands upon Kant’s notion that “intuitions without concepts are blind” to argue that nothing is “given” to awareness that is not already structured for awareness. Vasubandhu would agree, at least as it applies to ordinary sentient beings. See below.

75. Monier-Williams 1993:372; Tubb and Boose 2007:225–226; and Cardona 1967–68:320.

76. For a useful introduction to “intentionality” as a philosophical term, see Jacob 2010. If it seems unlikely that a Buddhist scholastic would have come up with this concept, recall that Brentano’s term is based on a medieval European scholastic one.

77. This is perhaps even broader than Searle 1992. This is not the place to attempt to locate Vasubandhu’s position in the forest of contemporary views on “intentionality.”

78. Searle (1992:20): “Because mental phenomena are essentially connected with consciousness, and because consciousness is essentially subjective, it follows that the ontology of the mental is essentially a first-person ontology. Mental states are always somebody’s mental states. There is always a ‘first person,’ an ‘I,’ that has these mental states.”

79. Recall Vasubandhu’s point that the distinction between what the eye “sees” and what the mind “sees” does not represent an ontological difference. There need be no unique ontology to the first-personal perspective if it is just a story that is told “about” an eye-consciousness—a tag with which it is labeled—when its image is reconstituted within a mental consciousness.

80. See Sharf 1998 for a parallel argument regarding religious experience.

81. The Buddhist view is that we are “programmed” by our karma to experience the world as we do.

82. It would appear that Bhāviveka was the first to launch a Mahāyāna critique of Yogācāra, in order to establish the Madhyamaka tradition as dominant—a doxographic project that took some time to gain traction, but did so, eventually, in Tibet. See Eckel 2008 on Bhāviveka and Vose 2009 for a study of the story in Tibet.

83. Garfield 2002:176: “External objects are nonexistent. But their conceptual construction by the mind is real. That construction, being itself purely mental, is not dual: it does not resolve into subject and object simply because there is nothing to be found on the object side—there is only the conceptual activity of the subject, which is mistaken for an independent object.”

84. TSN 156.21–24: cittamātropalambhena jñeyārthānupalambhatā | jñeyārthānupalambhena syāc cittānupalambhatā || (36) dvayor anupalambhena dharmadhātūpalambhatā | dharmadhātūpalambhena syād vibhutvopalambhatā || (37).

85. This argument also echoes the Viṃś, in which Vasubandhu argues against the viability of anything that might be called the real “grasped” objects of perception. The argument in the Viṃś is causal as well: since there can be no sensible account of the causal relation between physical and mental events, the traditional Buddhist understanding of sensation must fail. It follows from that argument, furthermore, that the unreality of perceptual objects leads to the unreality of the perceiving consciousness.

86. With regard to nirvāṇa, the term upalambha is clearly metaphorical, used for poetic assonance; there is no possibility of sensory grasping of knowledge by an omniscient Buddha.

87. Although the notion of “the emptiness of emptiness” as expressed in Madhyamaka is a unique structure, the self-referential, self-undermining logic of which it is an instance appears across Mahāyāna literature and is only one of many examples of philosophers’ attempts to address the “limit paradoxes” or “limit contradictions” that arise when we try to speak of the limits of thought and language. On the Madhyamaka doctrine of the “emptiness of emptiness,” see Huntington 1989, Garfield 1994, Arnold 2005, and Siderits and Katsura 2013. Mark Siderits (pers. comm.) thinks that my reading of TSN places Vasubandhu quite close to Candrakīrti. On limit paradoxes, see Priest 2002.

88. Viṃś 10.30: sarvathā sā tu na cintyā.

89. Viṃś 11.1: sarvaprakārā tu sā mādṛśaiś cintayituṃ na śakyate | tarkāviśayatvāt.

90. Viṃś 11.4–5: buddhānāṃ hi sā bhagavatāṃ sarvaprakāraṃ gocaraḥ sarvākārasarvajñeya jñānāvighātād iti.

91. VyY 239.15–16: chos rnams thams cad ming med cing || ming gis yongs su brjod pa yin.

92. VyY 239.18–22: brjod du med pa’i mtshan nyid dag chos rnams thams cad yin na ni | byis pa rnams kyis kun tu brtags pa bshin du brjod pas | rjod par byed pa de’i rjes so ’brangs nas ’phags pa rnams kyis kyang min med pa rnams ming gis yongs su brjod pa yin par rung ngo.

93. The discussion that follows refers to the central verses of the TSN, which I have provided with detailed, interpretive annotations in appendix G. The SNS is the source that describes emptiness with the terminology of “the lack of own-being of dharmas” (VIII.31 [86.1]: chos rnams kyi ngo bo nyid ma mchis pa).

94. See appendix G for a complete translation.

96. This paragraph refers to Vasubandhu’s commentary on MAV III.16c–d (44.15–21). D’Amato’s translation (2012:152–153) reads: “Form [i.e., the first of the five aggregates] is threefold. Form as it is imagined is the imagined nature of form. Form as it is conceptually discriminated is the dependent nature of form, because this is how the conceptual discrimination of form comes about [i.e., through dependence on causes and conditions]. The actual nature of form is the perfected nature of form [i.e., its emptiness or absence of subject–object duality]. And the same applies to the aggregates other than form, viz., feeling, [conceptualization, dispositions, and consciousness]; to the spheres; to the sense bases; etc. Thus the tenfold skillful reality is viewed in terms of the fundamental reality since the aggregates, etc., are included in the three natures.”

97. The great Madhyamaka philosopher Candrakīrti famously claimed that an ordinary cowherd is privy to all the worldly conventions (lokaprasidda) that can constitute genuine knowledge. Perhaps this claim was motivated by frustration with the well-known tendency within all academic communities to reify jargonistic terms and concepts. If the idea is taken too seriously, however, it denies the pragmatic benefits of expertise and theoretical framing. After all, not all cowherds affirm Buddhist doctrine or practice. See Cowherds 2011 for a contemporary discussion of the Madhyamaka attempt to untangle this issue.

98. Thus, although the frame is relativistic, it depends for its viability upon the ostensibly objective data available through its application to pragmatic goals. The ultimate “pragmatic” goal for Buddhists is, of course, liberation.

1. It may be argued that karma for Buddhists is not intended to be fair; the pathetic and unjustified suffering that results from its lawlike operation may be taken to be the very quintessence of the First Noble Truth. We ought to want to escape from saṃsāra, it may be said, precisely because it is unfair. Yet even so, karma is anything but arbitrary. Pleasant experiences result from well-intentioned, good deeds, and unpleasant experiences result from evil deeds. Karma is wonderfully “just” in the sense that the punishment always fits the crime. One of the strategies used to combat anger in both the Theravāda tradition of the Visuddhimagga and the Mahāyāna tradition of the Bodhicaryāvatāra is to recognize that one’s enemy will be punished whether or not one intervenes: if they’ve done something wrong, they have it coming, so you can relax and let karma take its course. Our main problem indicated by the First Noble Truth, then, is not the injustice of karma, but rather its tragic justice meted out when we act childishly, out of our ignorance and our inability to control our impulses. Perhaps the real injustice of karma is that it continually punishes as adults those who are at heart only minors. The Buddha, in his compassion, teaches us how to act our age and stay out of trouble. That is a far cry from saying that the whole system of samsara is corrupt.

2. See, for an introduction, Siderits 2007:69–84.

3. For a readable introduction to Abhidharma categories of mind, including an insightful analysis of the foundational importance of the kleṣas, see Dreyfus and Thompson 2007.

4. Under such a reading, Buddhism is broadly consequentialist, as Goodman (2009b) argues convincingly.

5. By “moral nihilism” I mean the view that everything of moral significance is an illusion—or, to say the same thing, that morality itself is unreal.

6. The sixth-century Madhyamaka philosopher Candrakīrti’s famous affirmation of the “emptiness of emptiness,” which undermines all claims to ultimate reality, including the claim that would undermine them all, is sometimes read as throwing one back into conventional reality. If ultimate reality is empty of ultimate reality, then there is no purchase to the “only” in what was previously taken to be “only” conventional reality. This perhaps reaffirms conventional reality, reality as we ordinarily understand it. It is difficult to reconcile such a view, however, with the Madhyamaka philosophers’ continued advocacy of practices such as philosophical analysis and meditation, which are clearly intended to liberate the mind from certain specific conventions. For an elaboration on this and related problems for Madhyamaka ethics, see Cowherds 2011.

7. Quoted in Patrul 1998:88.

8. This translation of the title is from Thanissaro 1997b. Further citations will be from Walshe 1995, abbreviated as SPS.

12. For an overview of Plato and Augustine on self-mastery, see O’Connor 2005.

13. As Meyers (2010:261) points out, however, free will and Buddhist liberation are not at all the same thing; the main freedom in consideration in the Nikāyas is not metaphysical “freedom of choice” but “freedom of control.” She also (2010:22) makes the interesting point that if we consider “freedom” to be the ability to choose among multiple courses of action, we might find the “liberated” being to be less free, in the sense of having to choose always the best action, the action that advances the path.

14. The word “damaged” is from Thanissaro 1997b. Walshe (1995a:547, n139) correctly comments that the passage literally says the king is “uprooted and destroyed,” indicating “that Ajātasattu was inhibited by his kamma from obtaining the results that would otherwise have accrued.”

16. A classic statement of this argument is Van Inwagen 1975.

17. For a summary of libertarian views of free will, see O’Connor 2002.

18. Siderits 2007:69–84.

19. Goodman 2009b:150; Griffiths 1991.

20. This clever translation is adopted by Meyers (2010).

21. The view I advocate, then, agrees in structure with where Meyers (2010) points out that a nondeterministic metaphysics is not required for one to believe in the ability to choose one action over another.

22. Meyers 2010:253: “An intending (cetanā) is a kind of exertion, a movement of the mind towards a particular object or end, but the fact that intendings are present in all states of consciousness, including those that we would consider ‘unconscious,’ means that an intending is not, in itself, indicative of choice or control and that karma is not always subject to free will.”

23. For this way of putting the point, I am indebted to Christopher Knapp (pers. comm.). The same conclusion is drawn by Meyers (2010).

24. A possible counterexample here might be the Ājīvikas, but given how little we know about their beliefs, I am inclined to assume that their determinism was similar to that of the Sāṃkhya, as described in the next paragraph: objects, but not subjects, are causally determined.

25. Miller (1986:151–152): “Arjuna, the lord resides / in the heart of all creatures, / making them reel magically, / as if a machine moved them. | (18.61) With your whole being, Arjuna, / take refuge in him alone— / from his grace you will attain / the eternal place that is peace. / (18.62) This knowledge I have taught / is more arcane than any mystery / consider it completely, / then act as you choose / (18.63).” See Malinar 2008 for an analysis of the complex Sāṃkhya theory that makes this an arcane mystery, not a self-contradiction.

26. An alternative, which does mesh in certain ways with the structure of the Mahābhārata, is that urging and convincing is an important part of the proper lead-in to warfare, and indeed any of a king’s important acts. My sense is that the urging has an important causal impetus, but only because it places before the king the reasoning and the motivation for making the right choice.

27. One reading of the notion of determinism operating at an “ultimate” level when free will operates at the “conventional” level would be to say that we are thrown “back” into the conventional once we see the impossibility of altering things at the ultimate level. This might be thought to counteract the utility of Abhidharma analysis, a view of Mahāyāna that Vasubandhu rejects in the VyY. See chapter 4.

28. Whitney 1885:47.

29. Though cetana as an adjective is not common in Buddhist texts.

30. “These ten dharmas are omnipresent in every mental moment.” AKBh 54.19: ime kila daśa dharmāḥ sarvatra cittakṣaṇe samagrā bhavanti.

31. AKBh 54.20: cetanā cittābhisaṃskāro manaskarma.

33. Rabten 1992:114. Batchelor translates cetanā here as “intention.”

34. AKBh IV.1b (192.9): cetanā tatkṛtaṃ ca tat.

35. Meyers 2010:174.

36. “Cetanā is that by which the course of action is brought to completion. Āśaya is its purpose/intention [abhiprāya], ‘I will do such and such, and I will not do such and such.’ ” AKBh 271.11–12: cetanā yayā karmapathaṃ niṣṭāpayati | āśayas tadabhiprāya evaṃ caivaṃ ca kuryām evaṃ caivaṃ ca na kariśyāmīti.

37. AKBh 192.23: cittavaśena kāyasya tathā tathā saṃsthānaṃ kāyavijñapti. Note that this definition applies only to “bodily informative action” (kāyavijñapti), not all bodily karma.

38. AKBh 195–196. This passage is complex, and although I believe I have fairly expressed an implication of the jaḍatvāt on 196.1, it is admittedly somewhat ambiguous.

39. At AKBh III.25a, Vasubandhu says that the official Vaibhāṣika position is that the true meaning of pratītyasamutpāda is that they represent twelve distinct states of the five aggregates (133.10: āvasthikaḥ kileṣṭo ’yaṃ). This verse is one of the instances, which we listed in chapter 3, note 19, where the verses use the term kila to indicate traditional Vaibhāṣika doctrines with which the author has some disagreement. According to La Vallée Poussin (1923–1925.2:66, n. 5), Vasubandhu’s Vaibhāṣika rival Saṃghabhadra attributes the alternate view here to “Le Sautrāntika,” meaning Vasubandhu. This is evidence of the unity of authorship for the verses and the commentary. (“Saṃghabhadra: Les maîtres d’Abhidharma disent que c’est en considérant les « états » (avasthās) que le Bouddha enseigne le Pratītyasamutpāda. Le Sautrāntika (=Vasubandhu) ne le croit pas, et c’est pourquoi il met dans sa stance le mot kila [que nous traduisons: « d’après l’Ecole »].”)

41. “The situation of the virtuous, etc., actions in the previous life is what they call conditioning [saṃskāra]—by whose action the result is here.” AKBh 131.22–23: pūrvajanmany eva yā puṇyādikarmāvasthā seha saṃskārā ity ucyante yasya karmaṇa iha vipākaḥ. The point seems to be that under this interpretation, we call previous actions “conditions” to the extent that they ripen into results here. Such a reading localizes the intentions of the Buddha to make a specific point, and reminds us that, for Vasubandhu (as against the Vaibhāṣikas), we should not expect to find a singular explanation of saṃskāra or any of the twelve links.

43. Pruden 1988–1990 has “consciousness” for bhava here: 2:407.

44. A corollary piece of evidence here is that a defilement cannot cause a thing. If the twelve links represented only blind, mechanistic causality, there would be no reason that defilements should only cause actions. The current structure suggests that a defilement is just understood as the cognitive and emotional content that prompts deluded actions.

45. An entire scritpural “basket” (pitaka) is dedicated to the monastic vows.

46. This is implicit in the Sāmaññaphalasutta when Ajātasattu is unsatisfied by his conversation with Pūraṇa Kassapa, who holds the Ājīvika view of “nonaction” (akarma) (SNS 93–94).

47. Vasubandhu uses the word paṅkti for a line of ants. This analogy is clearly what Śāntideva has in mind in his Bodhicaryāvatāra VII.101, as the commentator Prajñākaramiti understood. Williams’s suggestion (1998a:105, 113–114) that this refers to “a caste line” is therefore probably mistaken.

50. The path of the bodhisattva takes three “countless” eons, where an eon is one Brahma-lifetime, and the word “countless” is a not literally infinite, but a very large, finite number. Following Kapstein’s notes on Dudjom (1991.2:9–10, n109), I calculate Vasubandhu’s explanation of the term (AKBh III.93d–94a) to add up to 10⁵⁹.

51. Meyers 2010:174.

52. The idea that reasons and causes are two discrete regions of analysis, but that, contrary to Davidson, this does not prevent reduction, may be taken to be yet another expression of the Three Natures view: what Sellars calls the “logical space of reasons” is the fabricated nature; the “causes” are the dependent nature; and the fact that the “logical space of reasons” is only a construction, a delusion brought about by causes and conditions, is the perfected nature.

53. Swinburne (1991:76) considers it a satisfying “terminus for explanation” if we can provide a “personal explanation” of an event that includes a person’s ability and desire to act. We do not need to reach behind the desire and explain that, too. This is key to his proof of the existence of God, since he argues that God’s will and ability to make the world explains it, whereas no scientific explanation could ever be similarly complete. Vasubandhu would answer that every person’s desires are, by their nature, conditioned, so to ascribe desires to God would be to define God as conditioned. (See Gold 2011 for Vasubandhu’s disproof of the existence of God.) He might also add that it is crucial to the illusion of self that we do not question its apparent unity, in which it appears to us as though it were uncaused and eternal. Once we know that our apparent identity is causally conditioned, we must change our view of the nature of the self. It is, therefore, only delusion that makes Swinburne’s explanatory criterion appear “satisfying.”

54. This from Barack Obama’s famous July 13, 2012, campaign speech in Roanoke, VA, in which he was pointing out that people who start businesses rely upon others who have paid for road construction, education of workers, and so forth. Obama’s opponents criticized this acknowledgment of causal indebtedness as un-American.

55. AKBh IV.1 (192.5): karmaja lokavaicitryaṃ.

56. This is the secret moral of the opening line of every Jātaka tale, every story of the Buddha’s previous lives: “Once, the bodhisattva was a Brahmin living in Benares”; “Once, the bodhisattva was king of the monkeys”; “Once, the bodhisattva was in hell.” Contrary to our ordinary practice, the Buddhist tradition extends personal identity across lives. What is significant about us—what is valuable about us—is the set of moral choices of which we are capable, rather than our status within a particular body within the cycle of saṃsāra.

58. “Defensiveness” is Thanissaro’s translation (1995a). Walshe (1995b:564, n334) cites other translators’ choices as “watch and ward,” “protection,” and “safeguarding.”

1. This question is pursued by Siderits (2009).

2. Priest 2002: e.g., 252.

3. See chapter 4 and appendix E.

4. I do believe, however, that Vasubandhu’s position is a faithful synthesis of mainstream Buddhism for most of its history in India, the land of its birth. As mentioned in chapter 4, Indian Buddhist philosophy after Vasubandhu was dominated by the pramāṇa schools, which come under direct influence of his perspective, and the tantras, the practice of which explicitly manipulates conceptual constructions in a mind-only frame.

6. Here we might add that even this way of speaking of space—as a location in which objects may be said to exist—is mistaken from the perspective of general relativity.

7. Wegner 2002:263: “The experience of consciously willing an action is something that happens in a virtual agent, not in a brain or in a mind.” Westerhoff (2010) details a wide range of correlations between the best contemporary cognitive science and the elements in a traditional Buddhist list of examples illustrating the illusory nature of experience.

8. Here it is worth mentioning the oft-noted convergence between Buddhism and psychotherapy. What I would add to this well-studied issue is a convergence in the approach to causality and the mind. To put a fine point on it, the secular, psychological mode of description is Buddhist Causal Framing as defined above. Remember that the key for BCF is always to return to the core question of the causal story in which entities participate. It hardly needs mentioning that the common psychological story of the development of the personality, the person’s identity, challenges the unity of the person by recognizing the conditioning roles played by relationships with parents and siblings and teachers, for instance, in the formation of character. Not only does this recognition play a role in identifying the fact of, and the nature of, the identity’s construction, but it is also used to revisit those conditioning factors, reevaluate them, and manipulate their influence so as to transform the personality. What seems an intractable nature, what appears as a solid and indelible identity to the patient, is of course conditioned—and subject to further conditioning. This is, essentially, point 4: the self appears as an unchanging entity, acting independently and intentionally, but it is not, and cannot be, given what we know about its history and its development.

Furthermore, the way that the various sets of points flow—4, 5, and 6 from 1, 2, and 3, and 7, 8, and 9 from 4, 5, and 6—is readily exhibited in a psychological worldview. The recognition of the long, incremental and cumulative, conditioned history of the mind shapes the psychotherapeutic approach to changing it. Even if I am aware of the destructive patterns of my mind, I can only change them gradually, one moment at a time. Inclinations and tendencies are not transformed by individual decisions, but by repetitive exercises in transformative choices. I cannot decide not to be an addict; I can choose not to drink today. Or, to be more accurate, I can “decide” to not be an addict, but that decision is not sufficient to make me not an addict. Conditions trump intentions. Psychological thinking (especially behaviorist therapy) in this way adopts the perspective of Buddhist Causal Framing in its suggestion that our best rational choices, placed before us, provide causal conditions for our healing. But our behavior, even when we exhibit our strongest effort, is not “rational”; it does not exist as it appears. The self is a construction, and must be reconstructed through slow accruals of mental action.

9. Not every dictionary. For instance, not Johnson’s. Johnson’s dictionary (1755) defines “stone” as “a mineral not ductile or malleable” (324), which provides a different set of qualities, more legitimately supported by pragmatic, physical examination, than “solidity.” We are still left with the question of just what it is, though, that puts up resistance to the doctor’s foot, so we look up “mineral” and find “matter dug out of mines” (217). That is a rather pragmatic definition. So, we turn to “mine,” where we find, “a place where minerals are dug” (217). I do believe at rock bottom we have found a rainbow.

10. There are, of course, many capacities that have come about that are not advantageous. I assume that molecular-level vision would be expensive to maintain, and difficult to generate randomly, and so would not arise unless it conferred a significant advantage.

11. “Because it is under the influence of the ripening of their similar, own karma.” Viṃś 4.16–17: samānasvakarmavipākādhipatyāt.

1. Clearly it is past actions and their karmic results to which one can be “bound” and from which one can be “freed.” The pronoun “it” is probably intended to pick up the topic of the surrounding chapter, which is the “afflictive tendencies” (anuśaya) that bind one to samsara. The tatra tena vā, which I translate “in it or by it,” is repeated at the end of this passage, where the question is reiterated in reverse as “by it or in it,” tena tasmin vā.

2. Lit. “because they are bound to the quality of what is conditioned” (saṃskāralakṣaṇayogāt).

3. Here La Vallée Poussin (1923–1925.3:52–53, n2) imports the term anyonya from the description of the view into the list of names of views, making Buddhadeva a defender of anyonyathātva—a “mutual-differentialist.” That term works to describe this relativistic view, but the wording we have in both Sanskrit manuscripts, which is also reflected in the commentaries and Tibetan translations (pace LVP), is anyathānyathika, a “difference-differentialist.” What does this mean, exactly? Perhaps that there are different ways to differentiate between the different times, depending on your perspective. But perhaps the manuscript traditions for this passage have been corrupt since before they were translated into Tibetan, and LVP’s version, based on the Chinese implication of apekṣa, “relation,” is closer to the original.

4. Here Vasubandhu uses samanvāgama, which is his conventional replacement for prāpti in II.36 (as noted by LVP; see chapter 2, note 32).

5. The point is that whether the dharma is past or future, its relations to other dharmas are the same—so this fails to distinguish between dharmas.

6. Vasubandhu is applying his mastery of the complex Vaibhāṣika causal system. He refers to a potential cause, the tatsabhāga, which is not engaged in its causal activity, and asks of it what makes it present when it is present. According to the notion of kāritra on the table, a present entity must be engaged in some activity (“giving or taking a result”) to be called “present.” Even as a merely “potential” cause there must still be some activity that makes it present. So, if even a potential cause (tatsabhāga) has an activity, then an actual sabhāga cause must have two separate activities that make it present. That would be strange enough (already causal overdetermination), but the trick here, I gather, is that the way a sabhāga cause works—which is what the tatsabhāga potentially could have been—is that it “gives forth” its result when it is in the past. “Giving” causes are past conditions of present results (AKBh II.59). That means that the two activities for the same entity are at different times. If “activity” distinguishes the times, then this entity transcends the times.

7. A possible reading of this is that Vasubandhu is asking, “How can its own nature not exist yet or be yet-to-exist, if the thing itself exists?” An alternative translation would separate the passage into distinct phrases and take it as a word-by-word gloss: “[‘Existing in this way’ means that] the dharma is existing by its very own nature. [‘How can it be’ means,] How can this be established? [‘Unborn or destroyed’ refers to] what is called unborn, or is called destroyed.”

8. Is this comparable to TSN 4, dvayātmanā?

9. This is given as a quotation that the Vaibhāṣika accepts as authoritative.

10. Dan Arnold (pers. comm.) says that Vasubandhu is here adopting from Jain philosophers the idea that the existential verb acts as a nipāta. The point is that the word connects other words, rather than affirming some new entity.

11. Lit. “it has nonexistence.” The point here is that although you say this, you do not intend to say that the light has some reality as a nonexistent.

12. The “experiential object condition” is one of the four kinds of conditions. See AKBh II.61–62. Vasubandhu provides no technical definition for a “producer condition” (janaka-pratyaya), though this might be thought to correspond imprecisely to the “causal condition” (hetu-pratyaya), since at AKBh V.46 Vasubandhu argues that “producings” come about as a result of the assemblage of causal conditions.

13. Ājīvikas were a non-Buddhist heterodox (non-Hindu) school whom Buddhists generally depict as fatalists.

14. Here the discussion shifts to the topic of Buddhist doctrine. The Vaibhāṣika challenges Vasubandhu’s analysis by pointing out that there is no proper doctrinal structure for the mental perception of a nonexistent. On the contrary, the objects of the mental sense organ are dharmas, which are the entities of Abhidharma analysis. This challenge allows Vasubandhu immediately to turn the tables and show that for many examples of supposedly systematic thinking within the Vaibhāṣika, they must appeal to mere linguistic constructs, and not real dharmas, as mental objects—which is exactly what Vasubandhu believes.

15. The “one” here probably refers to the “consciousness” above—since the issue under discussion is the experiential object that is taken up by a consciousness.

16. This is rather a convoluted point, but as I take it Vasubandhu is making fun of the Vaibhāṣika’s apparent failure to distinguish between the sound of a word and its meaning, which ought to be the mental object for the mind that takes it up. Here Vasubandhu is saying that, according to their system, in order to call to mind a future sound, which does not yet exist (because it is future), one makes a sound (speaks a word).

17. That is, a real distinction could be asserted only if one says that one exists where the other does not. Since the Vaibhāṣika asserts that all things exist even in the future, this distinction cannot be made between future and present dharmas.

18. I have had some difficulty with this sentence.

19. That is, in AKBh IX. See the penultimate section of chapter 2 in this book for discussion.

20. If a thing is changed so that it comes into being, this is an acceptance of the expression that Vasubandhu is pushing here, that “not having existed it exists.”

21. That is, so the Sarvāstivāda tradition says.

22. This is Vasubandhu’s ironic statement of Vaibhāṣika literalism: yathā tu tad asti athoktam. It is clearly ironic because it follows two quite different interpretations of the same passage.

23. Afflictive tendencies (anuśaya), when activated, become defilements (kleṡa), which affect one’s experience, especially if one acts on them. The more general point is that one can give a causal account of being metaphorically “bound” by one’s actions in a way that prevents one from having to affirm that the past actions are literally doing the binding directly (and so must exist). See chapters 2 and 6 for my discussion of Vasubandhu’s view that karma works through mediated stages and the gradual accumulation of tendencies.

1. As Vasubandhu says when he cites this same passage in book IX, this is from the Paramārthaśūnyatā.

2. Here Śāstrī (2008) has an opposed reading: etad saty apy ātmani instead of etad asaty apy ātmani.

1. In the dhātu system, there are eighteen elements that make up all of the dharmas: six sensory objects, six sensory organs, and six consciousnesses. Here one sensory organ, the eye (cakṣu), and one sensory object, the mental objects (dharmadhātu), are classified as kinds of “view” (dṛṣṭi). Not all mental objects are views, though; eight kinds of mental objects are named, and the rest are said not to be views.

2. These first five are treated at AKBh V.7 (281.19–282.2). See chapter 3 for my translation and study of this passage.

3. The prajñā is one of the ten mahābhūmis, which are always-present aspects of mind.

4. Clearly this refers to the five consciousnesses that accompany the five sensory organs and their objects—that is, all of the consciousnesses except the mental consciousness.

5. This bold passage is my translation of one word, sabhāga, which is a technical term referring to dharmas that have an active (i.e., present tense) corresponding object. See AKBh I.39 (27.19–20).

6. The Vaibhāṣika argument will be that if the consciousness is what “sees,” it must be aware of every visual object—but that is clearly not what happens in visual consciousness. So this view must be rejected in order to make it possible that the visual consciousness is unaware of objects that we ordinarily cannot see.

7. It seems to me that there is no need to appeal to a Vijñānavāda explanation here, as Yaśomitra suggests (Śāstrī 2008:91–92), since the point is that the sensory organ provides the support (āśraya) for the consciousness—which is how the senses are said to operate at I.9c–d. But the argument is clearly tending toward a Vijñānavāda perspective.

8. Here we see Vasubandhu using the hermeneutical strategy of requiring consistency of purpose in texts, which McCrae (2000:433ff) writes about in the case of the Mīmāṃsā.

9. Vasubandhu rejects the question.

10. The context here makes it clear that conventional terms used “by choice” or “as one pleases” (chandataḥ) refer to the Buddha’s own figurative use of conventional terms.

1. This refers back to where the opponent had said that there can be no birth from a different class. If a variegated form arises from threads of individual colors, then this rule is violated.

2. The point seems to be that if one sees an individual thread, one must be able to distinguish different shades or colors—which means one sees a variegated form even when the thread is all the same color.

3. Given that fire’s radiance is so diverse and variegated, there is no unified “color” or “form” that could be said to be the single cause of the appearance of fire. We do not, therefore, “see” a fire; we see lots of separate things and construct the “fire” conceptually.

4. Things arise from multiple causes coming together.

5. A more literal reading here would be “There is no thought of them with respect to burnt wool, cotton, safflower, and saffron, etc.”

1. Based on the next passage, I am following Lee’s suggestion (2001:236, n1722) and leaving out the ma. With it, the passage would read: “the specific character of dharmas is not ultimate.” This might be taken to apply to the specific character of the dharmas designated by conventional language, but then how do we explain the subsequent critique of two kinds of existence applying to the same entity?

2. Here “those two” refers to karma and results.

3. Here the claim seems to be that of course a Buddha’s awareness is nothing but “means of knowledge,” but the only awareness that should count as “means of knowledge” for ordinary beings is the awareness that follows from the transcendent vision of Buddhahood.

4. This sentence introduces the discussion that continues to the end of this appendix.

5. Difficult to convey in English is the idea here, clearly expressed in the Tibetan (one imagines expressed even more clearly in the original, lost Sanskrit), that the dharmas have no existence in, or with reference to, the words. All of the singulars have been translated as plural to match English idiom. A literal translation of this line would be “It is not that it exists there” or “It is not that it has it.”

6. This sentence is different from SNS.

7. Here “Buddhahood” (sangs rgyas pa) is used verbally to indicate the substantial truth behind the language of the Āryas. I translate it as “awakened,” but it is clear that what one is awakened “to” or “into” is the inconceivable nature of Buddhahood.

1. Since the first quarter-verse clearly fits with the third and the second with the fourth, I have reordered the verse: a–c–b–d.

2. The verse uses the expression dvayātmanā, which I translate here as “as dual” to refer to how the illusory duality appears, and then uses advayadharmatā, which I translate here as “there being no duality,” to refer to the appearance’s fundamental lack of a false, dual nature. The alternation between ātmanā (lit. “with a nature”/“with a self”) and dharmatā (“true nature”) enlivens the dead metaphor of ātmanā.

3. Possibly “the activity of the seen, the seeing, and the cognition”—taking vitti separately from dṛg.

4. Here we begin the section that most evidently integrates the three natures and shows how they are paradoxically inconceivable.

5. From one perspective, we might see this as just expressing a general fact about a constructed error. It is real from its mistaken, constructed perspective, and unreal from the perspective of truth. But the grasping is not denied its reality. The verse does not say that there is real reality and unreal reality. It says that there are two ways to look at this entity—so it is both real and unreal. This means that the assignment of unreality alone is as much a mistake as the assignment of reality alone. If we take it from the SNS that the fabricated nature represents the doctrines of the Śrāvakayāna, this verse expresses that it is essential to employ, but not be attached to, those doctrines.

6. The dependent nature is the causal story of how the subconscious mind brings about the appearance of what appears to exist. This verse tells us that the causal story is real, but the appearance is not. This verse thus confirms that the causal story of the mind, which is the story of how we end up with the fabricated world as it appears, is opaque to us. Notice that if the Three Natures were providing a phenomenological description of the mind, we would expect to find out that the dependent nature does not appear at all. But in fact what the verse says is that the dependent nature appears under an error—it is as though error disguises the dependent nature, so it appears as something that is not there. But this is not the same thing as saying that the dependent does not appear at all. It appears. It is, in fact, all that appears, because there is nothing else that can appear. It is the only real, causal series. But it is only known through the existence of an error; how it appears is false. This verse therefore brings together the viewing-as an error with the viewing-of the mental causal story itself.

7. The perfected nature for the TSN is the absence of a real svabhāva where it appears to exist, in the fabricated nature. This definition weds the perfected to the fabricated—it is not as though the perfected is some additional reality, behind or above the fabricated. It is an inherent characteristic of the fabricated itself. The real-and-unreal character here is not, therefore, merely a logical problem of the form (–x) = –(x). One of the basic points of Yogācāra is the affirmation of the reality of nonexistence not as a thing, but as an appearance. We saw this affirmed in the opening verse of the MAV. Here the existence as nonduality is therefore affirmed as a thing with as much “reality” as anything else. Again, the point of there being both reality and unreality is affirmed as against a simple reduction of these two to a single description.

8. There are three quite interesting points here. First, note that construction takes place in a dual manner, or through the rule or application of two. The word vidhi means, centrally, a grammatical rule, so we may take it that fabrications are conceptual-linguistic dualities. Second, we have the idea that whereas reality is a dual construction, unreality is singular. We see this point in the next verse as well. What’s the point here? We may have a trick, an attempt to sound profound in the context of a need to find a way that this is unitary. More likely, though, is that the unreality of the fabricated nature is the perfected nature, the inconceivable ultimate that is always and everywhere the same—so, unified. This is affirming, then, that the perfected is one of the natures of the fabricated (its nature as unified). Third is the notion that the correct view includes both duality and unity: so, while we normally would say that Yogācāra holds to the view that all things are nondual, in fact the more appropriate point would be to say that things are beyond duality and unity. This verse thus launches the recursive logic of nonduality.

9. Here we have a similar point to the previous verse, where the unreality of the construction was said to be a “singular existent” (ekabhāva); here the fact of it being an error is said to be a “singular existent.” This is probably, again, a reference to the perfected nature, the ultimate, unitary emptiness. But we might also point out that the dependent nature is unitary here in the sense that it is a single causal stream that is only “dual” in how it is perceived. The error is itself “singular” even though it appears as “dual.” It is a unitary causal line that only appears as subject and object, action and agent. This is a central point that is expressed in the magical illusion analogy, encompassed in the notion that the duality is analogized to the elephant alone, not to the elephant-and-the-audience. The elephant’s reality is the “dual” nature of reality, the subject/object structure, which is generated (split) out of a single stream of awareness. See Gold 2006 for more on this point.

10. This verse is further confirmation that, for Vasubandhu, the perfected nature may not be taken to be the ultimate nature outside of, or over and above, the dependent and the fabricated natures. Clearly the three are intertwined. There is no way to speak of the perfected nature outside the duality of which it is the unitary nonexistence, and in which it exists. There is an alternative reading here that would make it “the nonexistence of duality” instead of “the existence of duality.” This is a somewhat more disappointing reading, since it makes the two perspectives essentially the same (-x) = -(x). Again, we have the point that nonduality is unitary.

11. Here Vasubandhu shifts gears, to talk about the conventional meanings of the three natures, which he calls a practical, or introductory (praveśa), perspective on them. The verses that follow provide, for each nature, its different stages of contemplation. These verses thus locate the “progressive model” of the three natures as a useful method of contemplation, which is nonetheless subordinate to the “pivot model,” wherein all three natures are descriptions of the same reality. These terms for different ways of interpreting the Three Natures are from Sponberg 1983; see also D’Amato 2012:15–16. I am only partly content with the idea of a “pivot” between fabricated and perfected with the dependent as the ground of both, since these verses place the nature of the perfected within the fabricated and vice versa. Perhaps the “crystalline model” or the “integrated model” would be better. But whatever term is used, the key for Vasubandhu is that all three are thoroughly integrated aspects of the same reality.