The present study deals with the epistemological problems of the scriptural authority in Mahāyāna Buddhism. It examines the various views on the valid source of knowledge (pramāna) in understanding of scriptural word in Mahāyāna traditions. The Buddha advises disciples to reject any form of transcendent authority. The Kālāma sutta teaches to reject the religious scriptures as a source of authority (pitika-sampadānena).1 Although this sutta focuses on the traditional Brahmanic religious texts, it also is true in Buddhism. On the surface, in the Buddhist scriptures there appears to be significant diversity and contradiction. Therefore, the Buddha advises that one should not accept scriptural authority without proper rational grounds, as the wise men test authenticity of gold by burning, cutting and rubbing on the touchstone.

Can Buddhist scriptures be sufficient grounds for the valid means of knowledge (pramāna)? Buddhists traditionally reject the validity of scriptural authority (sabda) as pramāna. For Buddhists, the Āgamas are not a perfect authority like in the Advaita Vedānta and the Mīmāmsa traditions. These two schools have traditionally the four sources of valid knowledge, namely, (a) sense perception, (b) inference, (c) intuition, and (d) revelation.

In the Mahāyāna Buddhist traditions, there are various views on the authority of Buddhist scriptures as a valid source of true knowledge. Buddhist logicians such as Dignāga (480–540 CE) and Dharmakīrti (600–60 CE) admit only two pramānas; perception (pratyaksa) and inference (anumāna). They regard the scriptures as a part of inference. For them, scriptures are not an independent pramāna, but they inform us only of the intention (vivaksa) of a preacher. They have no direct connection with reality.

On the other hand, the Svātantrika-Mādhyamikas, such as Bhāvaviveka (490–570 CE) insists that the words of the Buddha have epistemic validity. The Buddha gave privilege to knowledge, claiming that he speaks of what he knows and that others, too, can know what he himself knows. Bhāvaviveka emphasizes the role of reason (yukti) as a verifier of the authority of the scriptures. The function of reason is to establish a correct understanding of the scriptures. The scriptures assume validity only after examination by reason.

The Prāsangika-Mādhyamikas such as Candrakīrti do not agree with Dharmakīrti and Bhāvaviveka’s theory of scripture. In their views, the scriptures cannot, indeed need not, be pramāna at all. The Mādhyamika position is to deconstruct ontological or epistemological association of the scriptural word. As for the She-ling school’s view of pramāna, it shall be noted that, because at the time of Chi-tsang, the Buddhist logician’s works had not been introduced in China, therefore, the She-lingians might just follow Nāgārjuna’s theory of pramāna.2 It was Tsung-mi (780–841 CE), a synthesizer of meditation and doctrinal schools, who admitted the three sources of valid means of knowledge, i.e., inference, direct perception, and the scriptural word of the Buddha. Let us examine these positions in regard to the relationship of pramāna and the scriptures.

1The Kālāma sutta in the Anguttara nikāya (Hare 1973, 172, verse 65).

The Dignāga school is primarily concerned with the sources of valid knowledge and language. Since Dignāga admitted only perception and inference as the valid sources of knowledge, scriptural authority and revelation are not considered as independent means of proof. He defines perception (pratyaksha) as direct intuition, as distinct from indirect knowledge which comprises thought-construction, conception, judgment, and inference. It is pure sensibility, the very moment in the process of appreciation. Memory and imagination then build up sensory experience into a perception. Perception is thus the indispensable condition of all real and consistent knowledge (Conze 1967, 267; Stcherbatsky 1930, 146–61; 1932, 14–25). For the Buddhist logicians, perception is a direct or immediate cognition of an object that transcends all thought elements. They regard direct perception as a knowledge of the undifferentiated directly seeing, an entity as it is in itself.

On the other hand, inference (anumāna) is indirect knowledge which is mediated through conceptions and thought-construction. Reality is a series of moments, each moment is the sense-data, and each data, because it is pure particular, can not be said to be like or unlike anything else. Inference subjects particular entities to generalizations, thus, ending up knowing things not as they are in themselves but only as they are constructed by thought and language. Buddhist logicians try to distinguish valid from invalid inferences.

Following Dignāga, Dharmakīrti considers connecting the scriptures and the valid means of knowledge in chapter two of the pramānavārtika-svavrthih (Dharmakīrti 1938; 1943; 1959; 1960; 1964; 1971–72, 1–206). He points out that a word is not necessarily connected with an object.

Then, does the scriptural word have no real connection with reality? However, Dharmakīrti has to admit epistemic validity of the scriptural word, especially inferential validity. The scripture is an inference which informs the Buddha’s intention. Dharmakīrti asserts that the scripture is not a perfect authority which refers to true reality of beings. Even sage’s words are nothing but an inference. Hence, Dharmakīrti’s opponent will raise a question:

For Dharmakīrti, “a person in this world cannot live without depending on the authoritativeness of the sacred tradition. This is because one learns from Dignaga's view of agama that great blessing and great misfortune result. When one performs some acts, albeit only mentally, the consequences of which are not within the reach of our present experience. And because some conviction that agama is right can be derived from the fact that we do not see a contradiction to this statement regarding acts and their future consequences. So it will be better that one acts in such a manner as prescribed by āgama. By taking such practical reason into consideration, Dignāga said that āgama is a pramāna, i.e., an anumāna” (Dharmakirti 1987, 6, verse 213). Thus, Dharmakīrti admits āgama as a pramāna, for it has a practical function, considering peoples’ ordinary act under its guidance. Then, he offers criteria for the validity of the scripture.

Such a statement as is internally connected (sambaddha), describes a suitable means and speaks of real aims of man, is worth examining whether it is a pramāna, and a statement other than such is not. Here, the three points are presented: (a) āgama has internal coherence and connection, meaning that scriptural statements help each other to point to a single aim in spite of their diversity; (b) āgama describes an available means; and (c) āgama speaks of a real goal of people. The āgamas which meet the three conditions can be considered worth examining for its authoritativeness as inference. Otherwise they are inappropriate scriptures. And if scriptures meet the above three conditions, but they are found not to disagree (avisamvada) when examined, then, the person acting in accordance with such will be considered truly right.

Dharmakīrti further examines the meaning of “non-disagreement (avisamvada)” of the scriptures. The non-disagreement is the non-annulment by our actual perception and by the two kinds of inference, with regard to two kinds of things, seen and unseen, which are the objects of the author’s perception and inference.3 The word “non-annulment (abadhana) by perception,” for Dharmakīrti, means that those things which were considered as perceptible in scriptures are such i.e., are perceptible in fact, as we practically perceive them, like blue, etc., happiness and unhappiness, the observation of characteristics, greed etc., and cognition. In the same manner, the word “non-annulment by inference” is explained as those things which are considered in scriptures as objects of inference not dependent on a sacred tradition (āgama). And the non-annulment also means that those which could not be inferred by the author’s inference are such, i.e., are not inferable for us, too, like the ātman, etc (Dharmakirti 1987, 8). Thus, Dharmakīrti insists that the non-disagreement refers to the correctness (visuddhi) of all objects which perception and inference can determine, for there is no other possibility (āgatya) to explain it. Dharmakīrti formulates the logical order of correctness of the non-disagreement as follows:

As Dharmakīrti tries to prove, the āgama serves for the attainment of the human goals and so is pragmatically useful for the same purpose. The speaker of āgama being so credible will give us useful-true teachings with regard to an object beyond our cognition. Thus, āgama as an anumāna also is closely related to the problem of an omniscience of the speaker, the Buddha. Therefore, if we can prove this, all scriptural words of the Buddha assume epistemic validity.

Dharmakīrti also deals with this subject in the pramānavarttika, insisting that the Buddha is an omniscient being (Stcherbatsky 1932, 38–43). However, Dharmakīrti admits that the inferential validity of āgama is not absolute. There is a possibility of error, for words have no inseparable relation with the realities but merely with the intention of the speaker. Dignāga also agreed with this point:

Therefore, for Dharmakīrti the scriptural word cannot be an absolute authority. The scriptures consisting of words are only a right inferential cognition.

2Dignāga’s Nyāyamukha was translated by Hsuan-tsang (596–664 CE) into Chinese for the first time. For this reason Seungrang and the She-lingians might not know the Buddhist Logicians.  3The two kinds of inference refer to vastubalapravtta-anumana, such as inference of fire based on the perception of smoke and agamasrita-anumana inference basis on agama teaching.

What does the Mādhyamika have to say about the claims of the logicians’ that scriptural word has epistemic authority? First of all, let it be conceded that Buddhist logic, as all others, is not innocent. It has ontological commitment and functions as a tool through which to defend a particular world view. By refuting the metaphysics of the Sautrāntika-Yogācāra, the Mādhyamika is able to deconstruct epistemology and logic. The Mādhyamika thinks that the Buddhist logicians claim the theory of pramāna to establish the existence of knowledge of the independent being, such as particular mark (svalaksana).

In the Vigrahavyavartani, Nāgārjuna questions the possibility of ontological knowledge. He refutes the four pramānas of the realistic school, namely, perception (pratyaksa), inference (anumāana), verbal testimony (āgama or sabda), and analogy (upamana), as well as the objects to be established by pramānas. But here arises a question:

For Nāgārjuna, perception like all other entities, is empty. So also are inference, analogy, and verbal testimony. The person who apprehends the things through pramānas is also empty. Thus, there is no cognition of things, and a negation of the instinctive nature of things that are not apprehended is logically odd (Nāgārjuna 1978, 6–7).

In Nāgārjuna’s system, there are neither cognition nor the objects obtained through it. The pramānas cannot be established either intrinsically or extrinsically. The objects and the cognition of objects are dependent on each other, neither makes any sense in itself. Both of them are empty. Nāgārjuna observes that if pramāna are self-established (yadi svatas ca pramānasidhih), then the means of valid cognition are established independently of the object of true cognition, for self-establishment does not require another thing (Nāgārjuna 1978, 30). But the means of valid cognition requires the object to be cognized. Thus, pramāna cannot be established independently. Also, the pramāna are not established in relation to the prameyas, for the object to be established (sadhya) does not establish the instrument by which it is established (sadhana) (Nāgārjuna 1978, 31).

There is vicious circularity between cognition and objects. The circularity is stressed with the example of the son and father: “If the son is to be produced by the father, and if that father is to be produced by that very son, tell me which of these produces which other. In exactly the same manner you say; the prameyas are to be established by the pramānas, and those very pramānas in turn are to be established by those very pramānas. Now which of these are to be established for you by which others?” (Nāgārjuna 1978, 33). There will be neither son nor father, neither pramāna nor prameya, in this relationship.

Following Nāgārjuna, Candrakīrti also insists that Dignāga’s argument on the possibility of knowledge cannot be a valid means of attaining a true nature of thing, for it has already presupposed the realistic concept of self-nature as the object of knowledge. There is nothing to guarantee that the attainment of knowledge rests on the pramāna. Candrakīrti asserts that if the Buddhist logician cannot refute this objection, their correct definitions have no true explanatory power. Candrakīrti’s objections are:

The self-characterizing particular (svalaksana) is denied because a subject of characterization is unintelligible without actual characteristics. Following Nāgārjuna’s argument that if the subject of characterization is not established, characteristics become unintelligible, Candrakīrti argues that it is logically impossible for something to be characterized by itself because of the difference in meaning between means and end.

Although knowledge is itself instrumental and the svālaksana is an integral part of this, the subject cannot be itself the means of characterization. In the Madhyamakāvatāra and the Prasannapadā, Candrakīrti further argues with the Buddhist logicians about the theory of unmediated self-awareness (svasamvitti), i.e., the objectivity is assured of being integral to the object cognized for cognition is by means of unmediated self-awareness. Candrakīrti cannot admit this theory. Instead, he holds that there is no unmediated self-awareness as quoted from the Questions of Ratnacuda:

Thus, Candrakīrti criticizes the Buddhist logician’s ideas of perception. For Candrakīrti, the logicians accept the non-Buddhist views of the pramāna which are devoid of sense.5 The pramāna cannot give us knowledge of independent entities. But, the Buddhist logicians merely use the theory of pramāna in defense of their ontological thesis.

On the other hand, Bhāvaviveka, an exponent of the Svātantrika–Mādhyamika, insists that all the words of the Tathāgata have epistemic validity. He further asserts that there is the speakable ultimate truth (paryaya-paramārtha-satya), while Candrakīrti accepts only the unspeakable ultimate truth. For Bhāvaviveka, the Buddhist scriptures are the valid means of knowledge, and the scriptures can be understood only by logical reasoning.

The role of reason is important to discern the deep meaning of the scriptural words. However, reason cannot verify scriptures, but it only helps a proper understanding of scriptures. In the Mochizuki’s Buddhist dictionary, it is said:

Shotaro Iida criticizes Mochizuki’s translation and interpretation on the Bhāvaviveka’s Mādhyamakahrdaya, for yukti is not a verifier of the authority of āgama. Instead, he insists, “the unique feature about the verse is that its author, Bhāvaviveka, while keeping āgama as the initial and final authority, makes yukti an inseparable means to correct understanding of the deep meaning of āgama.” Thus, yukti has close relationship with a pramāna in understanding of the scriptural words. Bhāvaviveka points out that all utterances of the Tathāgata are our pramāna because these words are the statements of those who have perceived the true nature of things. Therefore, those who have discerning minds do not see any discrepancy or incoherence in the Tathāgata’s word.6 Bhāvaviveka puts faith in reason and encourages Buddhists to pursue logical reasoning to understand the scriptures properly.

Candrakīrti may agree with Bhāvaviveka’s stance that reason is the means to correctly understand the scriptures. Actually no Buddhist can be given to a blind acceptance of scriptural authority. Bhāvaviveka has a point in not giving reason the authority to determine the truth or falsity of scriptural assertions. In fact, Buddhist scriptures suspend such an evaluation. Bhāvaviveka simply wants to use reason as a hermeneutical tool, as a means to understand  scriptural meaning.

4Prasannapadā is abbreviated as PSP hereafter.  5For discussions on the Mādhyamika view of language, see Daye (1975, 77–96), Streng (1967), Gudmunsen (1977), Robinson (1967), Yadav (1992), Huntington (1983, 325–39), Williams (1980, 1–45).  6Bhāvaviveka, Mādhyamakahrdaya, vol. 9. 91-3-2, 3. as translated by S. Iida (1966, 82–83). Bhavaviveka’s theory of agama and yukti and a study of chapter 5 of the Madhyamakahrdaya are found in Yamaguchi (1964).

We concede that reason has a hermerneutical function in understanding the scriptures, especially in the case of the Mahāyāna scriptures. The hermeneutical functions and its limit of reason are observed by Shotaro Iida as follows:

A critical reasoning is required to distinguish the scriptures of definitive meaning and interpretable meaning. Furthermore, the role of reason is to realize the final meaning of the scriptures through inference. Bhāvaiveka, as seen in chapter two, extends the two levels of scriptures into the two levels of reality, i.e., worldly realty and ultimate realty. Candrakīrti also distinguishes the two levels of teaching, but for him, the two truths are not two sets of reality, but two types of discourse. Also ultimately all scriptures are neyārtha; they all point toward something beyond themselves.

Candrakīrti does not agree with Bhāvaviveka concerning the validity of knowledge of the scriptures. If all forms of sutra have epistemic validity, how can Bhāvaviveka explain non-Mādhyamika sutras such as Yogācāra scriptures which he criticized. By yukti one can grade scriptures nitārtha and neyārtha. However, reason alone is not an arbiter of the authority of scriptures.

Bhāvaviveka’s emphasis on pramāna, especially inferential reason, is criticized by the Prāsangikas. Bhāvaviveka as a Mādhyamika deviates from the school’s main point. He appropriates validity to pramāna and tries to employ inference to his position which is against Nāgārjuna’s doctrine. Candrakīrti will not say that all the words of the Buddha are pramāna, for pramāna bounds the scriptural word to ontological claims. But he may say that all Buddha-words are merely realization of path (mārga). This de-ontological and de-epistemological attitude toward the scriptures has become a common practice throughout history of the Prāsangika tradition.

PSP Prasannapadā 明句論. Trans. Mervyn Sprung. (Candrakirti 1979).