Paving the Great Way: Vasubandhu’s Unifying Buddhist Philosophy

THIS BOOK is a study of the philosophical work of Vasubandhu, a fourth/fifth-century Indian monk who was perhaps the greatest Buddhist philosopher after the Buddha.¹ Vasubandhu’s works are well known in Indian, Tibetan, and East Asian Buddhist traditions. From his time to this day, and without a break, his writings have been widely cited and commented upon, his arguments used and debated, and his accomplishments praised. He is a familiar figure in contemporary Buddhist studies as well. His works have been a constant topic of investigation and translation for more than a century—including being the subject of one of the signature accomplishments in the field, the copiously annotated French translation by Louis de La Vallée Poussin, from Xuanzang’s Chinese translation of Vasubandhu’s Abhidharmakośabhāsya, published between 1923 and 1931. Everyone knows Vasubandhu. What is remarkable, then, is that we do not, by now, know Vasubandhu very well.

We can list his attributed works, and recite his life story, but what exactly was his intellectual contribution? Unlike other Indian Buddhists of comparable stature—Nāgārjuna, Asaṅga, Dignāga, Dharmakīrti—he is uniquely difficult to provide with a tag line. The reason for this is that, as any student of Buddhism knows, Vasubandhu did not stick to one tradition. He was the greatest systematic thinker in the Yogācāra school of Mahāyāna thought, and he defended core Yogācāra doctrines with arguments that are still worth our consideration. Yet before that, he was the greatest advocate for the non-Mahāyāna, Sautrāntika interpretation of Abhidharma thought—a critic of extreme views advocated within the dominant non-Mahāyāna school of the time, the Kashmiri Vaibhāṣika. And even before that, he was the author of the most concise and comprehensive synthetic summary of the Vaibhāṣika tradition itself. Thus, he can be labeled with three distinct scholastic identities—Vaibhāṣika, Sautrāntika, and Yogācāra—none of which applies to the full body of his work. What’s more, his contribution to each was unique.

The differences between the distinct scholastic identities represented in Vasubandhu’s works are underlined in traditional versions of his biography extant in both Chinese and Tibetan.2 Vasubandhu, we are told, was the younger half-brother of Asaṅga, the great founder of the Yogācāra school. Asaṅga had meditated for twelve years in a cave before receiving a vision of the future Buddha, Maitreya, and subsequently learning core Yogācāra treatises from Maitreya. Vasubandhu, for his part, is said to have had questions about the details of the Abhidharma traditions, and so he traveled from his place of birth—Gandhāra—to Kashmir, to study with the great Vaibhāṣika teachers there. After mastering the Vaibhāṣika tradition and, according to some accounts, alienating himself from his peers by his too-critical questioning, Vasubandhu returned home and began teaching the Vaibhāṣika tradition. By doing so, he was apparently violating declared Kashmiri intellectual property rights, but the treatise he was composing to accompany his teachings—the Verses on the Treasury of Abhidharma (Abhidharmakośakārikā, AKK)—was so masterly that, although composed illegally, it was too useful not to be adopted and commented upon.³

Yet Vasubandhu’s own autocommentary, the Commentary on the Treasury of Abhidharma (Abhidharmakośabhāṣya, AKBh), was directly critical of numerous elements within the Vaibhāṣika system. To be fair, the AKBh is critical of everyone. Its single most sustained argument is in its last chapter, the “Refutation of the Self,” in which the main opponent is not the Vaibhāṣika, but the Vātsīputrīya (aka Pudgalavādin), advocate of an ineffable, existent “person.” This commentary takes off from the last verse of the AKK, and it may be read as though Vasubandhu, free of the requirement to stick to the Vaibhāṣika system adumbrated in the verses, is able to lay into his main rivals. Still, the majority of the work is a verse-by-verse commentary on the Vaibhāṣika system (sometimes term-by-term), and there are few stones left unturned. The commentary pits the Vaibhāṣika against a great many philosophical opponents, and the winners of the debates are not always clear. When they are clear, however, the victory often goes to a position named “Sautrāntika”—“those who follow scripture”—which is why this commentary is said to belong to the Sautrāntika school. This is a problematic term, however, because it appears that Vasubandhu was the first to use it as it is used in the AKBh.

Nonetheless, with these narratives in hand, Buddhist scholastics have an account of three distinct doctrinal identities into which to ascribe Vasubandhu’s diverse body of work. But once we are done explaining the three scholastic traditions in which he participated, there is nothing left to express as the intellectual identity of Vasubandhu himself, or the writings attributed to him. He is just a great scholar, who is able to master, and synthesize, everything he touches. His brilliance alone is what makes him great, as Paramārtha’s biography states:

Presumably, the point is that those who study the Hīnayāna use Vasubandhu’s Hīnayāna works, and those who study the Mahāyāna study his Mahāyāna works. This is the view implicit in the standard Tibetan doxographic categorization of Buddhist doctrine into four philosophical schools: three traditions represented by Vasubandhu’s career, capped off by Nāgārjuna’s Madhyamaka.⁵ When you want to study Vaibhāṣika and Sautrāntika, you look to the AKK/AKBh. When you want to study Yogācāra, you look to Vasubandhu’s Yogācāra works (along with those of Asaṅga, of course). This is a fulfillment of Paramārtha’s claim that Vasubandhu made good textbooks for the study of each tradition.

As a summary of Vasubandhu’s oeuvre, however, this is most unsatisfying. First of all, to suggest that Vasubandhu was merely setting out a doctrinal structure for easy digestion is to entirely ignore his unique contributions to each system, and his vibrant philosophical voice. Peter Skilling describes Vasubandhu’s prose style as “confident and learned, replete with citations and allusions to canonical and other literature, and to the opinions of different teachers or schools.”⁶ We may add, as I will argue throughout this book, that Vasubandhu repeatedly displays a preference for certain specific patterns of argument, certain methodologies in the application of reason and scripture, and remarkably—given the diversity of the topics covered—certain epistemological and ontological claims. These patterns within Vasubandhu’s works, I will argue, allow us a glimpse into what drives him as a thinker. Thus, whereas it is difficult to pin down Vasubandhu when we look at him from a doctrinal perspective, he is nonetheless readily identified by the light of his distinctive arguments.

What is missing, then, in the approach to Vasubandhu’s texts that divides his identity into his texts’ purported doxographic categories is an awareness of the continuity in Vasubandhu’s philosophical methods, his implicit presuppositions, his argumentative style, his patterns of reasoning, and his uses of scriptural citation and hermeneutics. When we seek this unity, we find that Vasubandhu’s texts are not so easily divided into doctrinal categories after all. And perhaps this is as we should expect, since our best recent evidence is telling us that “Sautrāntika” was not definitively attested as a doctrinal school before Vasubandhu, and “Yogācāra” definitely postdated him. The present study emphasizes the evident continuity of interest and purpose across diverse works attributed to Vasubandhu, and paints a picture of a great thinker’s central concerns and philosophical trajectory. It renews our understanding of Vasubandhu as one of the most important Buddhist philosophers of all time.

In the remainder of this chapter I will discuss the vexed question of whether there might be “two Vasubandhus,” and describe in more detail the methodology of the present work. The second and third chapters provide fresh, close readings of several passages from Vasubandhu’s AKBh. These careful readings are a workbook in Vasubandhu’s views and approach, introducing the reader to his characteristic arguments related to causality, time, perception, conceptual construction, ethical responsibility, and the language of Buddhist scripture. It is hoped that, together, they will provide the reader with the skills to identify the distinctive flavor and direction of Vasubandhu’s thought.7 For it is in the next two chapters that these themes are reintroduced as they appear in Vasubandhu’s “appearance only” (vijñāptimātra) works—chapter 4, focusing on epistemology and scriptural interpretation, and chapter 5, on perception and the nature of consciousness. These chapters argue that Vasubandhu’s unique contributions to (what would later be called) Yogācāra enshrine, in a doctrinal system, his unique preoccupations and contributions to non-Mahāyāna Abhidharma. The central chapters of the book also emphasize the many ways in which Vasubandhu’s arguments, and his purpose overall, are to be understood as serving a larger project of scriptural interpretation—a fact that is too rarely acknowledged and understood by modern interpreters of Buddhist philosophy. The final two chapters examine some of the significance of my characterization of Vasubandhu’s views for Buddhist philosophy more widely construed. Chapter 6 draws upon Vasubandhu’s view to clarify vexing issues in Buddhist ethics, in particular the nature of moral agency and the conundrum surrounding Buddhism’s apparent disinterest in talk of justice. Finally, the conclusion works toward a more formal statement of this view, which I term “Buddhist Causal Framing,” and shows how it may be deployed in arguments regarding idealism and the theorization of culture.

While I believe that Frauwallner’s thesis has been discredited, it remains of great importance for the light it sheds on the history of the discipline, and, more narrowly, the question of how best to study Vasubandhu. When I tell colleagues in the field that I am writing about Vasubandhu, the question often arises, “Which Vasubandhu?” Sometimes the questioner is asking with sincerity, and sometimes with tongue in cheek. But any scholar hoping to write about “Vasubandhu” must either elect to focus on one or the other of the “Vasubandhus,” or do battle with Frauwallner’s long shadow. Since this book seeks unity across works attributed to Vasubandhu, I have no choice but the latter option.

I am, however, by no means a pioneer in opposition to the “two Vasubandhus” view. From the moment of its publication, Frauwallner’s thesis was under fire. It has had its supporters, but almost immediately, the central significance of the argument—what it was proposed to solve—became difficult to sustain. Frauwallner had argued that the author of the AKK/AKBh was a different “Vasubandhu” from the brother of Asaṅga, and that Paramārtha (or Paramārtha’s students) had combined two different life stories in his famous “Life of Vasubandhu.”9 The AKK/AKBh was composed later, Frauwallner said, by another Vasubandhu, in the century after the Vasubandhu who was Asaṅga’s brother. The argument is a sophisticated one, and is primarily based upon a useful observation that Paramārtha’s biography is written out of chronological order, in a way that makes it seem like the story of the author of the AKBh was cut and pasted into that of the brother of Asaṅga. As Frauwallner points out, the story of Asaṅga’s brother seems also to have appeared elsewhere in Yogācāra circles independently of the story of the composition of the AKBh. Two separate narrative streams do seem to have been combined in Paramārtha’s biography of Vasubandhu, and Frauwallner’s thesis of Paramārtha’s error accounts for this.¹⁰

Yet the critiques that have arisen in response to Frauwallner’s famous thesis seem far more weighty than the problem of the combination of narratives. P. S. Jaini based his critique on a previously unstudied Sanskrit manuscript, the Abhidharmadīpa with its commentary, which includes critical references to the AKK/AKBh.¹¹ The author of this text criticizes the author of the AKBh for having adopted the Mahāyāna and the “Three Natures” view. Unless this evidence has been misread, it seems to justify Paramārtha’s “combination” of the AKBh narrative with the Yogācāra identity associated with the brother of Asaṅga. In a subsequent publication, Frauwallner himself clarified his view to agree that the author of the AKBh certainly wrote some Yogācāra texts, but he held onto the two Vasubandhus view as a view of two individuals, each of whom started as a Vaibhāṣika and switched to Mahāyāna.¹²

It seems to me that the discussion might well have ended at this point.¹³ It is difficult to see how the odd structure of Paramārtha’s biography provides any bulwark against a strong textual tradition affirming that the author of the AKBh adopted Yogācāra later on. Even if Paramārtha is assumed to have combined stories about his subject from different sources, he is hardly the first biographer to have done so. What’s more, the existence of distinct narratives that emphasize the AKBh and the Yogācāra identity, respectively, can easily be explained by the fact that Vasubandhu’s works were, in fact—according to the received narrative—inherited and taught in distinct institutional lines.

In the subsequent decades, a number of authors have sought to counter Frauwallner’s thesis—especially its initial version—by describing, sometimes in great detail, apparent continuities across Vasubandhu’s works. Lambert Schmithausen argued that two Yogācāra works, the Twenty Verses (Viṃśatikā, Viṃś) and the Thirty Verses (Triṃśikā, Triṃś), showed sufficient continuity with the AKBh to be considered as participating in the “Sautrāntika presuppositions” (Sautrāntika-Voraussetzungen) of a single author.14 Akira Hirakawa traced terminology and doctrinal structures from the AKBh to the Yogācāra works via intermediate texts such as the Analysis Establishing Karma (Karmasiddhiprakaraṇa, KSP) and the Analysis of the Five Aggregates (Pañcaskandhaprakaraṇa, PSP).¹⁵ Stefan Anacker did the same with closer attention to lines of argument.¹⁶ These are texts that clearly continue the analytical projects of the AKBh, but include, and entertain, terminology classically ascribed to the Yogācāra. Peter Skilling surveyed twelve texts attributed to Vasubandhu, arguing that they share sufficient stylistic commonalities and cross-references to be attributable to the same author.¹⁷ A number of other scholars have done similar work in reconstituting Vasubandhu’s oeuvre from various perspectives.¹⁸ In the last decade, Robert Kritzer has advanced the continuity thesis to its extreme, arguing that the AKBh was written under a covert Yogācāra influence.¹⁹ My own reading of Vasubandhu’s continuity is influenced by these authors, whose arguments I find largely convincing.

Yet all the king’s horsemen and all the king’s men have yet to put Vasubandhu together again. Schmithausen still allows that Frauwallner’s thesis of “two Vasubandhus” may apply to distinguish, not the author of the AKBh from the Viṃś and the Triṃś, but the commentaries on Maitreya texts. And Skilling sticks to his twelve texts, also refusing to commit on the Maitreya commentaries, the sūtra commentaries, and minor works. For his part, Jaini has not pushed the point any further than to say that the author of the AKBh must have written some Yogācāra works. Far from disproving the “two Vasubandhus” thesis, Jaini even went so far as to affirm his support for the view.²⁰ While I have no difficulty admitting that this or that text may be falsely attributed to Vasubandhu the author of the AKBh, we have yet to identify a text that is legitimately attributable to the “older” Vasubandhu.²¹ Yet he lives on, hovering in the background, and like the ghost he is, scares away many who would otherwise approach a synthetic analysis of Vasubandhu’s works.

The evidence that convinced Jaini was not Paramārtha’s biography, but rather another point treated in Frauwallner’s article: the fact that Yaśomitra, commentator on the AKBh, supposedly made the distinction between two Vasubandhus himself. Frauwallner cites a few, rare cases where Yaśomitra refers to vṛddhācāryavasubandhu and sthaviraḥ vasubandhu, which he says refers to “an older Vasubandhu.”²² As Anacker has argued, these titles could easily mean “eminent” rather than “old,” and these passages may be read to foreground the AKBh author’s own view amidst a diversity of cited options in a complex discussion.²³ Although I see no mystery here, and no need to accept that Yaśomitra knows of some “older” Vasubandhu, Anacker’s analysis has not satisfied all sensitive readers.²⁴

A more intricate knot is formed around a passage where Vasubandhu cites a particular approach to the antarābhava, the intermediary consciousness of a living being between death and rebirth, and attributes this approach to “pūrvācāryāḥ,” which Frauwallner translates as “teachers of yore.” Yaśomitra explains the phrase pūrvācāryāḥ with the gloss, “Yogācaras: Ārya Asaṅga, etc.”25 Based on this gloss, and another example of pūrvācāryāḥ used with reference to Asaṅga, Frauwallner concludes, “Asaṅga was therefore for [Yaśomitra] an old teacher in comparison with the author of the Abhidharmakośa” and that Yaśomitra believed that “Asaṅga belonged to a considerably earlier time than the author of the Abhidharmakośa.”²⁶ Given that Asaṅga was, by traditional accounts, Vasubandhu’s older brother, a great deal of unnecessary pressure is being put on the interpretation of the word pūrva, which means “previous.” The passage does not necessarily justify the term “old” in “old teacher,” but even if we take it that way, it could still refer to Vasubandhu’s older brother. Thus, Frauwallner’s “teachers of yore” for pūrvācāryāḥ is a selective translation that overstates his case. The phrase “a considerably earlier time” clinches the point, but only by inserting the word “considerably” without justification. Frauwallner’s is a sensible reading, but it is only one interpretation. Since Frauwallner will claim to be engaging in a decidedly “scientific discussion” of the topic, let us note for the sake of clarity that no “scientific” method is proposed for determining how long before Vasubandhu’s composition of the AKBh Yaśomitra intended the “previous” teacher Asaṅga to be stating the views in question. Yet it is upon the interpretation of Yaśomitra that the “two Vasubandhus” hypothesis stands.

Furthermore, Frauwallner mentions, and rejects, a moderating suggestion from Péri: the name Asaṅga might just have been used as an exemplification of “yogācāra,” since Asaṅga was the most eminent teacher of the Yogācāra school. If Péri is right, “previous teachers” could include Asaṅga’s Yogācāra teachers, who might be “older” still than Asaṅga, and thus much older than Vasubandhu. In response to this, Frauwallner writes, “To this we may oppose the fact that Asaṅga is the founder of the Yogācāra school.”²⁷ If we define the Yogācāra tradition as that which stems from Asaṅga’s Yogācārabhūmi, this “fact” will stand. We may even wish to agree that to identify Maitreya as the source of the school’s teachings is a pious fiction. But the question at issue is what Yaśomitra thought Vasubandhu meant by “previous teachers.” What is the evidence that Vasubandhu, or Yaśomitra, thought that Asaṅga invented his own view, and had no teachers? Given that the accepted narrative of the founding of the school includes Maitreya’s revelations, the scholarly community’s assessment of “fact” tells us nothing about what Yaśomitra believed, let alone what he thought Vasubandhu believed.

We might add, as a bit of helpful context, that every traditional Buddhist author claims to be passing on inherited teachings. I know of no evidence that Yaśomitra thought Vasubandhu thought Asaṅga had violated this norm. It would, absurdly, follow from Frauwallner’s reading that Yaśomitra must think that Vasubandhu thinks that Asaṅga himself made up the particular interpretation of antarābhava at issue in the passage in question. Even if the Yogācāra school can be traced back to Asaṅga, the views of the Yogācārabhūmi did not appear out of thin air.28 That every author is embedded within his or her context is a “fact” that is implicit in Buddhist doctrine, and that Vasubandhu never forgets.

Furthermore, as it stands, there is no reason to think of a “Yogācāra school” as a “fact” at all during Vasubandhu’s time—and this would apply to any Vasubandhu. Vasubandhu uses the term yogācāra (lit. “practitioner of yoga”) in only two instances in the AKBh, and in both cases he is clearly referring to practitioners of yogic meditative exercises, not to any philosophical school.²⁹ This textual fact alone suggests to me that Vasubandhu’s AKBh was most likely written before Asaṅga’s magnum opus, the Yogācārabhūmi, had had time to calcify into the name of a doctrinal school. The fact that Yaśomitra uses the name “Yogācāra” places him at a temporal distance after Asaṅga—a distance that is not evident in the AKBh.

If we return to Paramārtha’s text, then, under the assumptions of Frauwallner’s reading of Yaśomitra—if we ascend to the next round in the hermeneutic circle—we find a contradiction far more puzzling than the question of how to translate pūrvācāryāḥ. The AKBh is full of specific names for authors and viewpoints. If it was written a century after the Yogācārabhūmi, the flagship treatise of the Yogācāra school, why didn’t Vasubandhu use the term Yogācāra when referring to this tradition’s views? Perhaps even more confounding is the same point asked in reverse: why was Vasubandhu comfortable using the term yogācāra to refer to a practitioner of yoga, without evincing any concern that it might imply some scholastic identity? In addition, Frauwallner mentions in a note that Haribhadra, still later than Yaśomitra, includes both Asaṅga and Vasubandhu as pūrvācāryāḥ in the Yogācāra. If Yaśomitra knows that the AKBh was written long after the “Vasubandhu” who was a “teacher of yore” of the Yogācāra, why, we may sensibly ask, didn’t Yaśomitra include the vṛddhācāryavasubandhu in his gloss as well? These questions dissolve if we assume that there was one Vasubandhu, who was composing the AKBh at a time when the independent “school” of Yogācāra was still in the process of formation (i.e., the time of Asaṅga).

What would also require explanation, if Yaśomitra were clearly suggesting that there were two Vasubandhus, is the complete and utter lack of awareness of any important “elder Vasubandhu,” before the time of the AKBh, in the history of Buddhism, given that so many thousands of eminent scholars of Sanskrit, traditional and modern (to say nothing of Tibetans) have studied Yaśomitra’s commentary in the around fifteen centuries since its composition. I do not mean to suggest that there are no systematic misinterpretations of texts. Rather, I wish to call our attention to the fact that these passages, if they seem to clearly evince another Vasubandhu, only do so according to a newly imposed set of framing presuppositions. My point is not that every impartial observer would deny Frauwallner’s reading; it is that Frauwallner’s claim that his own view is one “impartially considered”—and that any similarly impartial reader would be “confronted with the fact” of Yaśomitra’s distinction between two different Vasubandhus—is mistaken. We are all conditioned by our interpretive presuppositions. The solution, to the degree that we have one, is to continually test our views for their adequacy to the fullness of the evidence, and to avoid placing too much weight on any given interpretation.

Although I have a strong view about the “two Vasubandhus” hypothesis, then, my goal here is not to defeat it as much as to defeat the hermeneutic methods that support it, and that it promotes. In particular, I’d like to counter the air of scientific objectivity that bolsters the argument on the one side, and closes off alternatives on the other. The historical analysis of medieval texts is an interpretive practice, which requires that we enter the hermeneutic circle—we must admit that our assumptions shape our conceptions, and we may move by degrees toward a more adequate truth, but we are fooling ourselves if we imagine that we have adopted the perspective of absolute objectivity.³² Given that there is a necessarily subjective element in all interpretation, our search for greater objectivity requires that we pay close attention to the work that is being accomplished by our background assumptions and our interpretive tools. In particular, we are endangering our objectivity when the frame of reference itself appropriates the work of interpretation. When we hold such considerations in mind, the two Vasubandhus thesis reveals itself as a neat but untrustworthy explanatory device. Because of its conceptual structure, it appears as justified based upon too little evidence, and, once in place, it feeds itself on unresolved complexities. It is, in this way, not entirely unlike a conspiracy theory.

The dating of Vasubandhu has been a complex and difficult task for scholars, for reasons that are well attested in Frauwallner’s article. A half-century elapsed between Vasubandhu’s death and the composition of his first extant biography. That biography was composed in China—or, at least, it is extant only in Chinese—so the distortions attending the composition of hagiographies in India are joined with those attending translations or transitions from Indian languages to Chinese, and from Indic to Chinese culture. The dates for Vasubandhu’s birth given in different Chinese biographical accounts differ among themselves. The years 900, 1000, and 1100 after the Buddha’s nirvāṇa all appear. These may represent different calculations of the date of this event, but even so there is no way to square them. As of Frauwallner’s writing, there were two major theses regarding Vasubandhu’s dating; these placed him, respectively, in the fourth and in the fifth century. A wide range of minor pieces of evidence was called to account on each side. The major problem for each view was how to account for the evidence adduced by its opponents. The “two Vasubandhus” thesis, conveniently (I am saying too conveniently), represents a fiat that accounts for all of the evidence.33

Once the basic argument is in place, Frauwallner’s article runs through a variety of problems that are, to his lights, solved by the assumption of two Vasubandhus. One I have already mentioned is the fact that Yogācāra sources tell the life story of the first Vasubandhu (the brother of Asaṅga), without the section on the composition of the AKBh. Frauwallner takes it that this is explained by the fact that it is simply the story of the earlier Vasubandhu, not the later. At first glance, this explanation proves satisfying. But as soon as we return to the Vasubandhu narratives passed on by Yogācāra teachers, and attempt to delve deeper into the meaning of these stories, making use of the frame of Frauwallner’s new assumptions, we find a host of new interpretive problems. For example, the life story of the brother of Asaṅga always includes the famous event of Vasubandhu’s conversion from Hīnayāna to Mahāyāna—it would have to, because that is what happens to Vasubandhu in connection with Asaṅga. The problem, then, is who Vasubandhu was before he converted, if he was not the great author of the AKBh.

Inconveniently, the Yogācāra sources are unanimous in expressing his prolific abundance. Vasubandhu is said in one tradition to have composed a work called Abhidharmasāra in six thousand verses, and according to another to have written “500 Hīnayāna works” before converting and writing his “500 Mahāyāna works.”34 Recall that the central purpose of Frauwallner’s article is to prove that this story is not about the author of the AKK/AKBh. Frauwallner knows that the Abhidharmasāra itself is the name given for a text attributed to Dharmaśrī, which was a major source for the AKK/AKBh.³⁵ Is it not at least worth considering that this Abhidharmasāra with six thousand verses could be taken as a mistaken transposition of the Abhidharmakośa, famed for its six hundred verses?³⁶ If not, one would at least expect some explanation as to why not. And, what happened to those five hundred Hīnayāna works (that are not the AKBh or its relatives) attributed to this earlier Vasubandhu? When Frauwallner notes, as explanation, that the Yogācāra author tells us that the Hīnayāna works have been forgotten, we may be justified in wondering what has happened to his “scientific” critical perspective. Of course the Yogācāra authors wish to supplant Vasubandhu’s Hīnayāna identity with his Mahāyāna one. But even they admit a vast corpus of material; is it “objective” to accept their narrative as is? Should we trust the Yogācāra authors when they tell us that the great mass of non-Mahāyāna works of this influential author have simply been forgotten?

The central hermeneutic criterion, which I would advocate to avoid this kind of problem, then, is an ongoing, increasing depth of interpretation. There is a surface-level, technical sensibleness to Frauwallner’s position—that the five hundred Hīnayāna works are lost, and Abhidharmasāra is a name different from Abhidharmakośa. Yet in order to accept such a view we must forego a deeper reading of the very texts we are claiming to interpret. The dramatic character of Vasubandhu’s conversion, which is central to all of the narratives, is based upon his prior accomplishment as a non-Mahāyāna scholar. If the earlier Vasubandhu didn’t write the AKBh, then he is not a known, let alone an important, non-Mahāyāna scholar. According to the evidence of Frauwallner’s paper, the only non-Mahāyāna accomplishment we know of for this earlier Vasubandhu was that he agreed with the author of the AKBh about three or four points cited by Yaśomitra. The point is that the conversion narrative itself clearly unites the AKBh with the Yogācāra texts. It is only the adoption of a lens that seeks to split the evidence in every case that would prevent such a reading from rising to the surface.³⁷

The influence of background assumptions shows itself further in how Frauwallner evaluates the significance of particular pieces of evidence. Paramārtha says that Vasubandhu, the author of the AKBh, taught the son, called Balāditya, of one king Vikramāditya. The identification of these kings is crucial to Frauwallner’s argument, since we have a fairly good sense of the dating of the Gupta kings. If early Gupta kings can be found to match these names, then this attribution would place the author of the AKBh in the earlier period, the time of the brother of Asaṅga. But these names appear, for Frauwallner, as solid evidence connecting Vasubandhu the author of the AKBh with two later figures in the Gupta royal line, properly ordered, and known to be so named: Skandagupta and his nephew Narasiṃhagupta. While other scholars have noted that we know of another Vikramāditya, and hypothesized that other, earlier Gupta rulers might also have been called Balāditya in their youth, Frauwallner cites scientific authority to preclude such assumptions: “A scientific discussion cannot base itself on what name could have been borne by a king, but it must find out who actually bore it.”38

Of course, assumptions can be dangerous, and all will agree that we should not place too much weight on any given assumption. Yet this very strident comment appears only one paragraph after Frauwallner’s claim that “no difficulty is presented” to his interpretation by the admitted fact that there was, in truth, no direct succession from Skandagupta to Narasiṃhagupta, as his interpretation of Paramārtha’s text should require. For, he says, the intervening reign of Puragupta Prakāśāditya was so short as to “easily have been forgotten after 50 years,” when Paramārtha would have recorded the story. Science allows the assumption that our main authoritative source is wrong in just this way about the line of succession, but does not allow the assumption that we may be missing a name for one of the kings? This is selective leniency, affirmed under a cloak of scientific objectivity. Once we admit that the text’s transmission of the royal successions is imperfect, there arise numerous potential explanations of the evidence due to “forgettings”; perhaps the compiler of the biography knew that Vasubandhu taught the son of Vikramāditya, and asked his companion, “Who was it who succeeded Vikramāditya, again?”—and his companion gave him a name for the successor of the wrong Vikramāditya. In such a reading, we end up thinking there are two Vasubandhus because there were two Vikramādityas, and Vasubandhu worked for one of them. I am not saying that this necessarily happened, but it seems just as likely as the specific forgetting that is necessary for Frauwallner’s argument.

Later, Frauwallner proposes “another fact which goes against a connection between Vasubandhu and the earlier Gupta rulers,” which is that Paramārtha refers to the Gupta capital city as Ayodhyā, whereas the capital at the earlier time might well have been Pāṭaliputra. Frauwallner’s account of the dating of the Gupta capitals is by no means certain, and even if it were, couldn’t this, too, have been “forgotten” after fifty years—or lost track of in Paramārtha’s account? Such evidence is ambiguous. Why not admit that it is ambiguous? There seems to be a disconcerting pattern of claiming “science” when one’s own evidence reflects a particular kind of precision—this king was definitely named “Balāditya” whereas we’re not sure about the other king—and claiming “fact” when there is ambiguity, but only for favorable positions.

Perhaps the most convincing piece of evidence for the early composition of the AKBh, and the most damning expression of Frauwallner’s bias, is a verse from the Kāvyālaṃkāravṛtti by the grammarian and poetics scholar Vāmana. Vāmana adduces a verse with the phrase “refuge of the wise” (āśrayaḥ kṛtadhiyām) referring to “Candragupta’s son,” as a poetic example of “special intention,” or, in Frauwallner’s translation, “covert allusion” (abhiprāya).39 The allusion here, Vāmana says, is to Vasubandhu’s ministership. The wise one ministering to the son of Candragupta is Vasubandhu. This is clearly saying, from a source outside the lines of Vasubandhu’s heirs, that Vasubandhu taught the son of one Candragupta. Unfortunately, since Frauwallner entertains this piece of evidence only after he has already concluded that Paramārtha must be referring to the kings Skandagupta and Narasiṃhagupta, he says that Vāmana is in contradiction to Paramārtha. But surely the goal should be to make the best sense of all of the evidence. It may be that if Vāmana is accepted as evidence, this will require us to go back to the drawing board with our interpretation of Paramārtha. But that difficulty does not mean that it is one or the other.

Nonetheless, Frauwallner pits the evidence from Vāmana against the evidence from Paramārtha, and he proposes two reasons, and a rule of “sound criticism,” to prefer Paramārtha over Vāmana. First, he points out that Paramārtha is earlier than Vāmana, by some two hundred fifty years, though the verse upon which Vāmana comments is, he admits, possibly earlier than Paramārtha. Then, he provides an analysis of the general trustworthiness of each text’s historical evidence:

Vasubandhu stands for Paramārtha at the centre of his interest; it is his life, about which he writes. Vāmana writes a manual of poetics, in which he inserts an occasional remark on Vasubandhu, the origin of which remains unknown. We know very well what we should expect from Indian commentators in such cases. Now and then some valuable tradition, but side to side with it the worst examples of superficiality and often the purest nonsense. Under such circumstances we are compelled to say, that according to the rules of sound criticism the authority of Vāmana has no weight at all in comparison to Paramārtha’s.⁴⁰

Here Frauwallner’s self-contradiction rises to the surface. He admits the possibility of “now and then some valuable tradition” accompanying a verse, which in this case affirms the possibility that Vāmana’s reading represents a genuine interpretive tradition reaching back to the time of the Gupta succession in question. Yet this acknowledged possibility is not pursued. But why not? It is clearly not “purest nonsense.” What evidence is there that Vāmana, court poet to the eighth-century Kashmiri king Jayāpīḍa, would have chosen a random, “superficial” name as comment to a verse about the education of princes? At the very least, we have with Vāmana a translocal, Indian, courtly affirmation of Vasubandhu’s role at the court of someone named “Candragupta.” Quite possibly, we have a continuous line of commentary extending back to Candragupta’s time. The fact that such a reading correlates with Paramārtha’s contention that Vasubandhu was court-appointed instructor for a Gupta prince ought to confirm the substance of this comment. In such a light it becomes an extremely powerful piece of evidence.

Yet such a confirmation is unavailable, when to confirm the verse would be to contradict details already decided about Paramārtha’s history, of which it has already been said that “we may trust in it a high degree, since Paramārtha, according to the results reached up to this point, was separated from Vasubandhu the younger by a comparatively small space of time.”41 In combination with an unjustified tarring of the historical utility of Indian commentarial traditions, this authorization of Paramārtha is leveraged to invalidate Vāmana’s comment. Apparently, excepting where it skips one king in a line of succession and, of course, where it unifies two Vasubandhus into one, Paramārtha’s history is deemed trustworthy, even to the point that it may be used to thoroughly disqualify otherwise solid counterevidence. Yet, in truth, the thinnest thread distinguishes Paramārtha from Vāmana—the “unscientific” possibility that the name Balāditya might have been used to refer to one of Candragupta’s sons.⁴² Lacking such evidence of any given son of a king having been called Balāditya, and given that Paramārtha’s purpose is to write about Vasubandhu, and not about a whole bunch of verses, “the authority of Vāmana has no weight at all in comparison to Paramārtha’s.” I find it extremely strange that Frauwallner takes it as significant support of the objective quality of Paramārtha as a historical source, that “Vasubandhu stands for Paramārtha at the centre of his interest.”⁴³ The “rules of sound criticism” and “scientific discussion” here are called upon to justify, above all, the linear construction of the argument: first, validate Paramārtha, then use Paramārtha to invalidate Vāmana. A properly circular hermeneutic would double back and engage the possibility of a correlation between Vāmana and Paramārtha.

Anacker advocates the suggestion, from Le Manh That, that the title “Balāditya” was likely not a name for a particular king, but rather a title attributed to a crown prince who has been granted partial empowerment before his father’s death and his full ascension to power.44 This possibility would allow us to read Vāmana’s verse and Paramārtha’s text as both legitimately referring to Candragupta II, who was famously known as Vikramāditya, and most likely his son Govindagupta. Anacker’s thesis nicely solves the Ayodhyā problem as well, but there is no need to rehash that argument here. Florin Deleanu, more recently, has followed up on these arguments and found them somewhat wanting in certainty—but without dismissing them entirely. He provides another series of conjectures that might work as “very fragile missing links” establishing the earlier dating for king Balāditya.⁴⁵ Many of the pieces fall into place if we accept, with Deleanu, a dating of around 350–430 CE. Not everything. But notice that we do not need certainty to escape from the “two Vasubandhus” hypothesis; it was certainty that was keeping us trapped within it. With our net less tightly woven, we can cast it more widely, and seek out Vasubandhu, the author, wherever he may range.

The “two Vasubandhus” view’s vestigial survival, then, must be attributed to some degree to the resonant authority of its author, and to its comforting utility as a hermeneutic structure. Frauwallner surely deserves his scholarly reputation, but this hermeneutic is undeservedly satisfying and self-justifying. No doubt unintentionally, the view echoes the traditional Buddhist doctrinal compartmentalization of Vasubandhu’s works. Ironically, the “two Vasubandhus” thesis, while galling to scholastically trained Buddhist intellectuals, confirms their traditional doxographic categorization, according to which Vasubandhu is declared to have switched sides. Both approaches to Vasubandhu’s works reify the apparent divide between the text traditions of the “Sautrāntika” and the “Yogācāra” attributed to him, and both positions explain why Vasubandhu may have been separately valorized by Abhidharma and Yogācāra followers, respectively. Yet this parallel must itself be acknowledged as a methodological red flag.46

The scholars whose work upon which I am building here have undermined the strict division between Vasubandhu’s traditional doxographic identities. Traditionally, Vasubandhu was a Sarvāstivādin, then a Sautrāntika, then a Yogācāra. But there is no easy classificatory category, for instance, for the KSP; must we call it “Yogācāra” and not “Sautrāntika” just because it includes mention of the “storehouse consciousness” (ālayavijñāna)? To do so is anachronistic, as anyone will admit, since during Vasubandhu’s time, as I have said, the word “Yogācāra” had not yet taken on its distinctive scholastic meaning. It may be said that, though Vasubandhu never used the term to name his own view, he was a crucial player in the formation of what we now call Yogācāra scholasticism. So there is reason to take it as proto-Yogācāra. Yet I believe it is a mistake to set up a teleological directionality for Vasubandhu’s works, with the Yogācāra system as its ultimate end. The Yogācāra is surely where his doctrinal work culminates, but his own approach and vision within Yogācāra was itself a unique expression of his core values and philosophical interests. There is no satisfactory doctrinal category for Vasubandhu’s view, because his ideas cut across, and against, doctrinal identities. Even the AKBh, whose core philosophical position may safely be labeled “Sautrāntika,” collects within it countless alternative viewpoints, only some of which are refuted. When we wish to characterize his view, we must call it Vasubandhu’s Sautrāntika, or Vasubandhu’s Yogācāra, so as to distinguish it from standard doxographic identities. But given the continuities, let us just call the view “Vasubandhu’s.”