Discussion

Devadatta is fat but never eats by day: is Mīmāṃsā “postulation” just inference to the best explanation?

January 16, 2026 · 5 replies · 195 views

Reason & experience American Indian Modern European Mimamsa Nyaya C. S. Peirce Dignāga Gautama (Akṣapāda)Kumārila Bhaṭṭa #arthapatti #inference-to-the-best-explanation #postulation #pramana

When I teach sources of knowledge comparatively, I like to point out that the European and Indian debates ask different questions. The European one asks which source does the real work, senses or reason. The Indian one asks how many irreducible sources there are, and argues about whether each can be derived from another. That second question turns up things the first barely notices, and my favourite is arthāpatti, usually translated "postulation" or "presumption".

The textbook case: Devadatta is fat, yet he never eats during the day. Therefore he eats at night. The Bhāṭṭa Mīmāṃsakas, following Kumārila, counted this as a distinct means of knowledge, alongside perception, inference, comparison, testimony and non-apprehension. The Naiyāyikas, who counted four, said it was really a kind of inference. Dignāga's school, with only two, would agree with the Naiyāyikas on that point at least.

A reader trained in analytic philosophy shrugs: obviously this is inference to the best explanation, the thing detectives and scientists do every day. But I think the shrug is too quick. The mark of arthāpatti is that the conclusion is the only way to make two established facts consistent; the technical phrase means roughly "it could not be otherwise". Inference to the best explanation is comparative and probabilistic: of the available hypotheses, this one explains best. Those are not obviously the same operation.

So my questions: is postulation a species of inference, or something else? And does the Indian insistence on counting it separately reveal anything the European tradition missed, or did the Europeans simply file it under a broader heading?

Like ReplyFollow5 replies

Members reply here. Reading is always free.

5 replies

Govinda Prasad Mishra 9 months ago

Fellow

I teach this every week, so let me give the Bhāṭṭa argument. Inference needs a previously known invariable relation between a mark and what it indicates, as between smoke and fire. In postulation there is no such prior relation between "fat and not eating by day" and "eating at night". There are two known facts that appear to conflict, and the mind supplies the one thing that removes the conflict. The cognition arises from the conflict, not from a remembered pattern. That is why Kumārila would not let it be absorbed into inference.

Helpful · 3
Meera Kulkarni 9 months ago

Fellow

The Nyāya reply is that the relation is there if you look for it: wherever someone is fat and does not eat by day, they eat at night, and the absence of night-eating would mean the absence of fatness. That is a negative form of inference, and Nyāya was happy to count it as such. So the dispute is partly about how strictly you define inference. The Bhāṭṭas define it narrowly and need a new category; the Naiyāyikas define it broadly and don't.
Derek Hollis 9 months ago

Fellow

Peirce is the Western thinker who would have enjoyed this argument most. He insisted that abduction, reasoning to an explanatory hypothesis, is a third form of inference, irreducible to deduction or induction. So he made the same move as the Bhāṭṭas, recognising a distinct operation, but kept it inside the family of inference. Meera's point applies to him too: whether you need a new source depends on how wide you have drawn the old ones.

Helpful · 1
1. Kenneth Leung Chi-ho 9 months ago
  
  Fellow
  
  Derek, that's illuminating. Perhaps the honest conclusion is that the Indian count is partly a dispute about the boundaries of a single word, anumāna, which in the Bhāṭṭa usage is narrower than our "inference". But I still think the "could not be otherwise" mark is worth keeping. Inference to the best explanation never claims that.
  
  Helpful · 2
Emeka Nwosu 9 months ago

Fellow

A physicist's example. Astronomers measured how fast stars orbit in galaxies and found them moving too fast for the visible mass to hold them. Two well-established facts that seem to conflict. The proposed reconciliation is unseen mass, dark matter. That has the exact shape of Devadatta eating at night. But no physicist I know would say "it could not be otherwise"; alternative theories of gravity are still argued over. So in practice science seems to use the form of arthāpatti with the modesty of inference to the best explanation.

Helpful · 3