SuttaCentral

Epistemology, an ancient-modern debate and early Buddhism

Seeing as how Bhikkhu Sujato has been working on a series of lectures on early Buddhist epistemology, I felt this was the perfect time to post a thread regarding a new publication that came out just this year which outlines a long running debate in Tibetan Buddhist philosophy and to see what you guys think early Buddhists (or dare I say the historical Buddha!) would have said about this issue. If early Buddhist studies is to have applications beyond itself, then this is a chance to see how it might respond to later Buddhist debates and discussions, so let me outline the problematic.

The recent publication I am referring is Yakherds (2021). Knowing illusion: Bringing a Tibetan Debate into Contemporary Discourse. Yakherds is a group of scholars who mainly deal with Madhyamaka and Tibetan Buddhist philosophy, and they include Jay Garfield, a well known Western philosopher, translator and interpreter of Buddhist philosophy.

Garfield has a talk which outlines the issue well in this video (skip to minute 55).

He also outlines the main point of the debate in a paper which can be found on Garfield’s website. Garfield begins discussing how, in India and then Tibet, there were two main approaches to epistemology:

Transcendental epistemology which pursues epistemology by

“thinking analytically about the nature of justification itself, bracketing actual human practices, and develop a purely prescriptive account of epistemic warrant embodying a set of standards that might or might not be met by actual human practice.”

Then there is the second approach, the anthropological approach. According to Garfield

“On this approach, we begin not with analytical reflection on the meanings of epistemic terms, but rather by asking what actual people do when they claim to be justifying statements, or when they certify statements by others as warranted. In this approach to epistemology, we bracket questions about whether those practices meet some transcendental standard, and develop a purely descriptive account of actual epistemic practices. On this approach, we take what counts as knowledge to be a social or institutional affair, like what counts as currency or as a legal vote, and then ask about the institutional conditions on bestowing that honorific on a cognitive or linguistic episode. We might, for instance, discover that those in some community count as knowledge only that which is endorsed by scientists, or that another community includes the deliverances of certain oracles as knowledge. If we take this approach, we take it for granted that there is knowledge, and ask only what leads us to classify some statements under that head.”

Garfield notes how there are close parallels between these approaches and modern analytic philosophical positions (mainly, those of Carnap, Quine and Sellars). Carnap attempted to build a foundational epistemology based on sense data, like the Indian Buddhist pramāṇavāda thinkers Dignaga and Dharmakirti who defended two main ways of attaining knowledge: perception/pratyakṣa and inference/anumāna, as foundations of knowledge and held that they were ultimately founded on immediate sensory experience that puts us in direct contact with phenomenal particulars (svalakṣana). The macroscopic phenomena we encounter are mental constructs made up inferentially by the mind out of these particulars.

Garfield then explains that the position of Candrakīrti follows a more anthropological epistemology in his Clear Words (Prasannapadā). He disagrees with the pramāṇavādins in limiting the pramāṇas to just two (since this is not the only way people gain knowledge in practice) and he also rejects their foundationalism. Instead, according to Garfield, Candrakīrti “follows Nāgārjuna in taking the pramāṇas to be vindicated by the objects they deliver (prameyas) and by one another in a coherentist epistemology.”

However, Candrakīrti’s epistemological view can be interpreted in different ways and thus Garfield moves on to explains how Tibetan interpreters of Candrakīrti also differed in how to understand that way of doing epistemology (and how this is reflected by debates in modern analytic philosophy).

Garfield focuses on two major figures, Tsongkhapa and Takstang, and notes that they mirror in some key ways the responses of WV Quine (1908-2000) and Wilfrid Sellars (1912-1989) to Carnap (both of which defend a kind of anti-foundationalism with regards to knowledge, that is to say, that they both hold that knowledge cannot be grounded in a single thing, even so called sense data). However their positions differ quite radically. Quine thinks that since there is no absolute way to ground epistemology on any single set of epistemic primitives or foundations, then normative epistemology is impossible. Thus, while we can ask, descriptively, what it is that people do when they are doing epistemology, we cannot ultimately say whether they are right or wrong. This is thus a radically descriptive position which makes no normative claims (i.e. it does not say whether something is wrong or right, it just describes the practice of knowledge making as a human endeavor).

Meanwhile, Sellars’ position is different. Sellars argues that we can have a normative epistemology that is anti-foundationalist. That is to say, we can have secure knowledge about the world without relying on any single fixed point of reference or infallible epistemic grounding (contra Carnap who saw raw sense experience as in some sense foundational). As Garfield explains,

“he argues that neither meaning nor knowledge require foundations: meaning is constituted simply in the network of practices that constitute language use, practices that themselves induce the norms that govern syntax, semantics, and pragmatics; knowledge is that which is achieved by the appropriate use of epistemic conventions that themselves are justified by the knowledge they enable. He thus argues that we can make perfect sense of the norms that induce linguistic meaning as well as those that govern epistemic activity by attending to the power of conventions to induce normativity. Sellars argues instead that empirical knowledge has no foundation (or as Wittgenstein puts a similar point so perfectly: “the foundations are held up by the walls of the house.” We become knowers, on his account, when we come to participate competently in the collective social practice of justification and criticism; knowledge is just what we as a community of knowers take to be justified by our conventions of justification; no primitive world-experience relations are needed in order to constitute knowledge. ”

Just like how Quine and Sellars are both anti-foundationalists in radically different ways, Tsongkhapa and Takstang are also radically different Madhyamikas. As Garfield explains:

“They each reject the Dharmakīrti’s foundationalism, as well as his individualism. But while Tsongkhapa argues that this is consistent with a robust normative epistemology and the possibility of expressing a true Prāsaṅgika Madhyamaka position, Taktsang argues that at most it leaves us with an anthropological account of people’s deluded epistemic practices, and reduces Madhyamaka to inexpressibility. We hence see Tsongkhapa as following Sellars’ approach to the Carnap in his response to Digṅāga, and Takstang following Quine’s approach in his very different response.”

Let us briefly outline Tsongkhapa’s position (which, it should be noted, is the founder of the Gelug school, the tradition of the Dalai Lama). Tsongkhapa’s view on pramana follows Candrakīrti and Nāgārjuna in arguing that there is no foundation to epistemology. Instead, “the pramāṇas and their prameyas [what is known] are mutually dependent, and that the various pramāṇas also are mutually supportive, like the sheaves in a stack.” Thus, Tsongkhapa does think that he can do normative epistemology as a non-foundationalist because as Garfield explains

“ what generates epistemic warrant is not direct contact with things as they are, but rather a role in ordinary activity and confirmation by other warrants and they objects they deliver….We thus end up with an epistemology grounded in interdependence and in collective epistemic activity. Warrant arises not from direct, nondeceptive access to reality as it is, but from participation in a set of conventions that are mutually supportive and that constitute conventional justification. ”

This ultimately boils down to how Tsongkhapa interprets the key Sanskrit term samvṛti. As Garfield explains:

“Candrakīrti famously notes that samvṛti can either mean conventional in all of its familiar senses, indicating by agreement, ordinary, nominal, everyday, etc… or concealing, obscuring. So, we can gloss samvṛti-satya accurately either as conventional truth or as concealing or obscurational truth. One’s attitude towards the status of conventional truth depends a good deal on which of these readings one takes to be primary. Tsongkhapa takes the first route, emphasizing that to be conventionally true is a way of being true, not a sham that conceals the truth. He leans hard on Nāgārjuna’s doctrine of two truths, arguing that there can only be two of each of them is in fact a kind of truth. And since truth, or reality, and validation by pramāṇas are coextensive terms in this tradition, conventional truth is very much a way of being real, a way of being true, of being trustworthy. This is why Tsongkhapa can argue that the basis of division of the two truths is objects of knowledge: conventional and ultimate are each objects of knowledge, each real. We thus see a tight connection between normativity and reality: what is real is what is warranted by normative practices.”

Taktsang Lotsawa (a scholar of the Sakya school) could not be more different in his interpretation of Candrakīrti’s anti-foundationalist madhyamaka philosophy. For Taktsang, Tsonkhapa’s importation of pramāṇa and epistemology into Madhyamaka is mistaken, since the Madhyamaka view on ultimate truth is ultimate inexpressible and beyond all concepts and reasoning. Taktsang thinks that any commitment to epistemic warrants presupposes a foundationalist framework, and thus must be rejected. In other words, if there is no foundation for epistemology, no absolute fixed standard for deciding what is right or wrong, then how can there be right or wrong at all in some real sense?

Taktsang concludes that there is no possibility of a normative epistemology. We can only describe what people say about knowledge and justification in the everyday world (like anthropologists), but we cannot ultimately say that one justification is truer or more normative than another. Thus, for Taktsang, conventional truth is ultimately only “a truth for fools” and thus ultimately false. There can be no real knowledge of conventional reality (true knowledge is only ultimate, non-conceptual and inexpressible). He thus relies more on the deceptive and concealing aspect of the meaning of the term samvṛti.

As Garfield explains, these two Tibetan views on the middle way philosophy are closely mirror by the Western analytical philosophers we discussed above:

“Taktsang and Quine took the negative route, conceding that any account of knowledge and meaning that is genuinely normative must be transcendent, and so rejecting the possibility of a normative epistemology and of linguistic meaning, settling for a merely anthropological account of epistemic and linguistic practice. Tsongkhapa and Sellars took the positive route, arguing that convention could—and indeed must—ground normativity, and so arguing for a naturalistic but normative account of knowledge and of meaning that is conventional and coherentist, not foundationalist in character. So, while there is agreement among the principals in each of these debates that no transcendent account of normativity is possible, there is substantial disagreement about whether this dooms the search for normativity tout court.”

Garfield sides with the Tsongkhapa/Sellars side of the debate and proceeds to give some reasons about why. I strongly recommend reading the paper for those who are interested.

Now that I have outlined the ideas in the book and paper I mentioned (if you made it this far…wow, awesome), I want to get to the question I have been asking myself: how would the Buddha and early Buddhists make sense of this debate and which side would they fall on?

I think that, given K.N. Jayatilleke’s analysis of early Buddhist theory of knowledge, the Buddha (and early Buddhists) would agree with Tsongkhapa and Sellars and state that while there is no fixed and unchanging reference or epistemological foundation we can rely on, we can still have positive knowledge about conventional everyday reality as well as about spiritual matters. The extremely skeptical and negative epistemological stance of the Taktsang - Quine folks seems not to allow for making proper judgements regarding ethics, scientific truth, and spiritual matters.

I am particularly interested in what Bhante @sujato would say about this topic (as well as others on this board of course!).

1 Like

Firstly, both of them are quite far from the EBT: The Doctrinal distance between them is nearly negligible compared to the distance between EBT and any of the two. They all accept the Two Truths doctrine, the two selflessnesses, and so on. Even considering that, I do not agree with your statement that the early Buddhists would have preferred Tsongkhapa to Taktsang (and other critics of Tsongkhapa), at least in some important points. As you point out, one would find some similarities, but also dissimilarities as well.

For example, unlike the other Madhyamika (and almost all the other Mahayana schools), Tsongkhapa and Gelug scholars claim that to become arahant, the śrāvaka must realize both selflessness of pudgala and selflessness of dharma as the Bodhisattvas of Mahayana do. This implies that the teachings of “Hinayana” are insufficient to free from Samsara! This might sound unsuitable, but I think that it is impossible to extract sole epistemology from soteriology in the Buddhist tradition. (And there are important epistemological points beneath the rationale behind the claim of Tsongkhapa, as always.)

Such a series of comparisons will be endless. Thus, I would represent how Tsongkhapa and Mipham, as a representative of “the earlier schools,” i.e., Tibetan Madhyamaka schools other than Gelugpa, interpret the tetralemma of Nagarjuna differently. Note that in the 3rd lecture of Jayatilleke’s EBTK course by Bhante Sujato, almost the same tetralemma is explained. I just quote corresponding paragraphs of the translator’s introduction of The Wisdom Chapter: Jamgön Mipham’s Commentary on the Ninth Chapter of The Way of the Bodhisattva, translated by the Padmakara translation group.

A third important difference that divides Tsongkhapa’s Madhyamaka presentation from that of the earlier schools concerns the ultimate truth itself: its nature and the way it is to be realized. This question is articulated specifically in relation to the refutation of the so-called tetralemma, the four ontological extremes spoken of in Nāgārjuna’s kārikās. These four extremes may be understood in the following way. Phenomena are not (1) existent; they are not (2) nonexistent; they are not (3) both existent and nonexistent; and they are not (4) neither existent nor nonexistent.

This is the general introduction of the tetralemma of Madhyamaka. Now the Mipham and earlier schools’ interpretation on the tetralemma represents:

According to the interpretation of the earlier school, an interpretation with which Mipham is in full agreement, the purpose of the systematic analysis of the tetralemma is to show that the true status of phenomena is not to be found in any of these four ontological extremes.
Even though phenomena constantly and inescapably appear to us, when they are examined as to their actual nature, it is discovered that their status is logically inexplicable. Whatever that status may be—and we cannot ignore the fact that things continue undeniably to appear—it is something that the mind is powerless either to conceive of or express. Moreover, the arrival at such an understanding is not an ordinary discovery. It in no way resembles the finding of some hitherto unknown object in ordinary experience. Passing through the different stages of analysis associated with each of the four extremes, the ordinary mind—the mind of everyday experience—is itself brought to the limit of its powers of intellection, which must then be transcended. At that point, it is said, the mind becomes completely still, resting in a state that is completely free of conceptual elaboration (spros bral). It is a state that Chandrakīrti describes as a condition of utter silence. Even though, on the basis of a common experience of consciousness and knowing, this state is beyond the power of ordinary beings to imagine, the freedom from conceptual elaboration—so it is said—is very far from being a state of unconsciousness. It is the wisdom of the Āryas, the noble beings who have passed beyond samsara. (…)
Furthermore, it is said that this wisdom lies beyond the duality of subject and object, the duality that radically characterizes the states of ordinary knowledge.

Long story short, Mipham and the early schools highlight non-dual wisdom beyond the logic of tetralemma. Now it is Tsongkhapa’s turn.

Tsongkhapa asserts that the realization of the ultimate truth of phenomena must first proceed from the correct identification of the object of negation—the true existence of phenomena, as distinct from phenomena themselves. (…)
The goal of the Madhyamaka analysis is therefore a mental state of perfect conviction in the nonfinding of the object of negation. When, after long examination, this state of profound certainty is reached, the practitioner suspends the analytical investigation and rests his or her mind in the state of nonfinding. When the state of absorption dissipates, the previous analysis is resumed; and this alternating procedure is repeated again and again. (…)
In other words, for Tsongkhapa, the realization of emptiness does not imply a fundamental difference between the mind that achieves this state and the mind that, earlier on, had been engaged in intellectual analysis.

To understand the terms above in detail, one has to catch the idea of Tsongkhapa, like “non-affirming negative”. That might be out of the scope of my knowledge, thus let us move on to his interpretation on the tetralemma:

The refutation of inherent existence resulting in a nonaffirming negation is not, in Tsongkhapa’s system, followed by the refutation of the other ontological extremes. For him, the rules of logic show that the procedure of successive refutation described in the tetralemma is not to be taken literally. According to the law of the excluded middle, the denial of one term automatically involves the affirmation of its opposite; there is no third, or middle, possibility. Consequently one may begin by denying the true existence of phenomena, but if one then goes on to say that phenomena are not nonexistent, one is implicitly affirming that phenomena exist.

So far not so exceptional. But here comes Tsongkhapa’s ingenious twist.

When, therefore, Nāgārjuna speaks about the first two refutations, Tsongkhapa interprets him, not as implying some mysterious third affirmation, a state beyond existence and nonexistence, but simply as meaning, first, that phenomena do not exist on the ultimate level and, second, that they are not deprived of existence on the conventional level. The same procedure is then applied by implication to the third and fourth extremes. For Tsongkhapa, in other words, the tetralemma is not an intellectual yoga intended to push the mind beyond the confines of habitual intellection and into a meditative state free from mental proliferation. Instead it is a pedagogical tool for clarifying one’s understanding of the relationship between the two truths and for arriving at a correct idea of emptiness.

Thus, Tsongkhapa interprets that the meaning of the first lemma is for the ultimate truth, while that of the second one is for the conventional truth. It seems that Tsongkhapa might interpret similarly as Jayatilleke does in some points, but the extraneousness of his radical turn into the Two Truths is hard to be unnoticed.

Why does it happen? It is well known that the main quest of Tsongkhapa’s intellectual journey was to “save the conventional” by power of logic in the searching for the emptiness of phenomena: The emptiness of a cup must not eradicate the conventional existence of the cup in our daily life. His logical answer to this is that what Madhymaka negates in the searching of the emptiness is the true existence beyond the phenomena, the “real cup-ness” of the cup, not the conventional phenomena, the cup itself. Thus, he found a way to make the ultimate emptiness of the object and the conventional object compatible.

Instead, for Tsongkhapa, the two truths are mutually entailing and of equal importance. For him, in other words, the doctrine of the two truths is not a skillful means devised to wean beings away from their clinging to phenomena, in order to lead them to the realization of the ultimate truth. On the contrary, the doctrine of the two truths actually corresponds to the constitution of phenomena. (…)

Of course, his critics including Taktsang and Mipham were not happy about it. From the point of their view, what Tsongkhapa did was essentially a substantialist turn, since what remains after all the logical examinations is that a mere cup without cup-ness beneath. Instead, they claimed that the conventional phenomena and the ultimate emptiness are indivisible. Surely, logic can help ordinary people to clarify the situation. However, in the end, only prajna with the profound yogic experience, lets one realize the indivisible emptiness/phenomena as it is: The teaching of the Two truths is also skillful means, and will be overcome by such wisdom on the true nature of the phenomenon, according to Mipham and the other critics.

In Mipham’s view, however, it undermines what for him is the entire soteriological purpose of the teachings on emptiness, which is to free beings from the clinging to phenomena that is the cause of their suffering. Beings perceive phenomena; they do not perceive the true existence of phenomena. And since, on this basis, they are attached to things themselves, not to their true existence, it stands to reason that it is things, and not some putative true existence of things, that must be negated. For Mipham, therefore, to say that phenomena are not empty of themselves but only of their true existence is an important mistake, and he energetically refutes it on many occasions.

Thus, unlike Tsongkhapa who had to preserve the Two truth doctrine to the end, Mipham and the critics might get more points for their sublation of the Two truth dichotomy, which is alien to EBT. Furthermore, it is notable that while Mipham and the earlier schools follow the interpretation and theory of traditional Indian Mahayana scholars who would have encountered many contemporary “Hinayanists”, Tsongkhapa’s interpretation is unprecedented in Tibet or India.

However, this does not mean that Taktsang and Mipham are uniformly closer to the earlier Buddhists, compared to Tsongkhapa. For meaningful and constructive discussion, one should discover roots and relatives of the Two truth doctrines and non-duality in the EBT. Among the two interpretations, which might be acquiesced by nun Vajira? What would be the responses of the Tibetan Madhyamaka scholars, if they would have read the classification of the grasping aggregates (upādānakkhandhā) by nun Dhammadinnā? According to Bhante Sujato, the notion of the “object” of consciousness is absent in the EBT. Is such an absence kindred to the non-duality praised by some Tibetan Madhyamakas like Mipham?

1 Like

Hi, it took me a while to respond to this, since it is a long post and I have not had the time.

I disagree that Taktsang and Tsongkhapa’s views on madhyamaka are quite far from the EBTs, they are all ultimately discussing the view of Nagarjuna, which is closely grounded in mainstream Buddhist teaching. The two selflessness are not that radically different from what the Buddha teaches in the EBTs, one can find the idea that the person is empty and that the various dharmas, like the aggregates, are also empty (see Phena sutta etc).

Also, the idea that Tsongkhapa thinks that to become an arhat you need to realize both forms of selflessness does not make it farther from the EBT, since for Tsongkhapa, realization of one kind of selflessness entails the other. So really, this is not a problem. Indeed, Tsongkhapa’s view is closer to the EBT here, because other Tibetans argue that arhats do not realize emptiness completely, while Tsongkhapa does.

Regarding the two truths, I think Tsongkhapa is also closer to the basic EBT concepts which this idea comes from (i.e. nītattha and neyyattha) here because, as Y. Karunadasa notes, in EBTs “no preferential value judgment is made between nītattha and neyyattha. All that is emphasized is that the two kinds of statement should not be confused” (Karunadasa, ‘‘Theravada Version of the Two Truths’’). Furthermore, Karunadasa notes how in Pali, the term sammuti just means human convention and does not have this connotation of an inferior truth or a falsehood (which is a gloss for the sanskrit samvrti). Thus, Tsongkhapa, in maintaining that the conventional truth has some reality or “mere existence” (i.e. a relational, conventional, dependent and contingent existence) is IMO closer to early Buddhism than those who interpret Madhyamaka a kind of global anti-realism which sees conventional-transactional truths as being “truth for fools”.

But this discussion is kind of besides the point I wanted to focus on in my OP. My main focus was on epistemology. In this sense, Tsongkhapa’s epistemology is also closet to EBT IMO. Why? Because as Jayatilleke notes, early Buddhist epistemology takes seriously the need for proper means of knowledge. He argues that early Buddhist epistemology is a kind of correspondence theory with elements of coherentism. This fits well with Tsongkhapa’s conventionalist realism regarding pramanas (epistemic instruments, means of valid knowledge).

However, the other Tibetan interpreters of Nagarjuna tend to be more anti-realist regarding pramana, in some cases even becoming skeptics or pseudo-skeptics (like Patsab for example), holding that conventional truth is just what is agreed upon by a majority of people. This makes it very difficult to be able to argue for truth on the conventional level and approaches a kind of relativism which I find problematic. Of course, all would agree that the ultimate wisdom which realizes emptiness/anatman is in some way non-conceptual (even Tsongkhapa agrees with this), however, the point is that there is also a conceptual kind of wisdom and that indeed, this is where we need to start (with a conventional kind of right view).

Indeed, sutras like Kaccana are more rare than sutras in which the Buddha focuses on the conventional right view of rebirth, karma, and ethics. Thus I think that Tsongkhapa’s focus on conventional things like ethics and reasoning refreshing, even if it is not as sexy as the tantric Great Perfection mysticism of Mipham et al.

1 Like