Innocent until proven false: Kumārila on why a cognition needs no certificate

In the pāṭhaśālā where I teach, boys of twelve learn to recite long passages of the Veda with every accent in place, and older students learn the Mīmāṃsā arguments for why that recitation matters. Visitors sometimes assume the second is merely a pious defence of the first. In fact the Mīmāṃsā school, the tradition of Vedic interpretation whose foundational text is the Mīmāṃsā Sūtra attributed to Jaimini, built one of the most careful epistemologies of the classical Indian world, and its central idea can be stated without any reference to scripture at all. That idea is svataḥ prāmāṇya, the intrinsic validity of cognition. In this essay I explain it, set it beside its rivals, and argue that it deserves a hearing in modern debates about trust.

The idea

Kumārila Bhaṭṭa, the seventh-century Mīmāṃsaka whose Ślokavārttika is the classic statement, asked a simple question. When a cognition arises, say I see a pot on the floor, what makes it valid, and how do I know that it is?

His answer was that validity belongs to a cognition by itself, arising from the very conditions that produce it. The cognition does not need a second cognition to certify it. A capacity that a thing does not have of itself, Kumārila argued, cannot be supplied to it by something else. Invalidity, by contrast, is extrinsic. A cognition is overturned only by something further: either a later cognition that contradicts it (I go closer and see that the "pot" is a shadow), or the discovery of a defect in its causes (my eyes were inflamed, the light was poor).

Think of it as a legal presumption. A cognition stands until it is shown to fail. The burden of proof falls on doubt, not on belief.

The rivals

The Naiyāyikas, the logicians of the school traced to Gautama (Akṣapāda), disagreed. On the developed Nyāya view, the validity of a cognition is ascertained extrinsically: we confirm that a perception of water was correct when acting on it succeeds, when we walk over and drink. A cognition is just a cognition until it has earned its standing.

Kumārila's reply was a regress argument. If every cognition needs another cognition to establish its validity, that second cognition needs a third, and so on without end, and nothing would ever be known. Somewhere a cognition must be accepted without a further certificate. If that is allowed anywhere, why not at the start?

Within Mīmāṃsā itself, Prabhākara, Kumārila's great rival, went further still. On his view every cognition is valid in what it presents. What we call error is not a false cognition but a failure to distinguish two true ones, such as a memory of silver and a perception of something shining on the beach. My students enjoy arguing about whether this dissolves error or only renames it.

The application to the Veda

Now the scriptural part. According to the Mīmāṃsā Sūtra, dharma is known through Vedic injunction, because what ought to be done for the sake of unseen results lies beyond perception and inference. But if the Veda is a source of knowledge, what guarantees it?

Here the doctrine of intrinsic validity does its work. If the Veda's statements are valid unless defeated, the only possible defeaters are contradiction by another means of knowledge, or a defect in the source. Statements about dharma concern what perception cannot reach, so perception cannot contradict them. And the usual defects in testimony are defects in a speaker: ignorance, carelessness, the intention to deceive. The Mīmāṃsakas held that the Veda is apauruṣeya, not composed by any person, human or divine. With no author, there are no authorial defects.

This is why Kumārila, unlike the later Naiyāyikas, argued against a creator god and against the possibility of an omniscient person. An authored scripture would inherit its author's credentials, and those would need establishing. An authorless one does not.

The objections

Our tradition requires that the opponent's view, the pūrvapakṣa, be stated first and fairly, so let me state the objections.

Buddhist and Nyāya critics asked how anyone could know that a text has no author. Failing to remember an author is not evidence that there was none; many works lose their authors' names. The Mīmāṃsaka reply was that the Veda has been handed down from teacher to student in an unbroken line without any composer ever being recorded, unlike other texts whose authors are remembered. Critics were not persuaded, and the argument continued for centuries.

A modern reader will add another objection: intrinsic validity seems to make us too credulous. If every cognition is innocent until proven guilty, will we not believe far too much?

My argument

I want to separate the two halves of Kumārila's position, because I think the first is stronger than outsiders usually realise, whatever one concludes about the second.

The general doctrine, that we are entitled to accept what our means of knowledge deliver unless we have a specific reason to doubt, is close to what the Scottish philosopher Thomas Reid defended in the eighteenth century, when he argued that we have a natural disposition to believe what others tell us, matched by a natural disposition to speak the truth, and that relying on these is not irrational. It is close, too, to what contemporary epistemologists call anti-reductionism about testimony. David Hume, by contrast, wanted testimony weighed against our experience of how often such reports prove true, a position nearer to the Nyāya demand for confirmation.

The regress argument seems to me decisive against the strongest form of the extrinsic view. No one could begin to know anything if every act of knowing had to wait for a certificate. What the extrinsic theorist can reasonably ask for is something else: that we remain alert to defeaters. Kumārila agrees. His two kinds of defeater, contradiction and defect in the source, are exactly the questions a careful person should ask of any report. The charge of credulity therefore misfires. Intrinsic validity is not a licence to believe anything; it is a rule about where the burden lies, joined to a precise account of what can overturn a belief.

Whether the doctrine supports Vedic authority is a separate question, and I would not ask a reader outside the tradition to accept that it does on the strength of one essay. But the doctrine should not be dismissed because of its most famous application. It answers a question every epistemology must face: what do you need before you are entitled to believe? Kumārila's answer, the cognition itself together with the absence of any reason to doubt it, remains one of the best we have.