The canonical example: Devadatta is fat. He does not eat during the day. The two observations together are incoherent without a third fact — Devadatta must eat at night. The Mimamsakas argue that the third fact is genuinely known, by its own cognitive route. There is no perception of it; there is no general rule from which it has been inferred. What forces the knowledge claim is the requirement that the observed facts cohere.
The Mimamsa tradition treats this as a distinct pramana — a means of valid knowing — separate from perception, inference, comparison, testimony, and non-perception. Its operational shape is specific: an observation generates a knowledge claim by demanding what is necessary for its own coherence. The unstated is implicated by the structure of the stated.
Two flavours are worked out in the tradition. Postulation from perception runs from observed facts that cannot stand alone (Devadatta’s case). Postulation from testimony runs from utterances that would be incoherent unless some unstated fact is in play. Both share the same logical signature: an observed fact, an inability to stand on its own, and a postulated fact called forth by the coherence requirement.
Where English Falls Short
The closest term is abduction (Peirce) — inference to the best explanation. Abduction is comparative: among the available explanations, pick the one that explains the observation best. Arthāpatti is constrained more sharply. It operates when the observation cannot stand on its own at all; the postulated fact is the one that makes the observation coherent, and ranking among alternatives is a separate question. Abduction asks “what is the most likely explanation”; this Mimamsa concept asks “what must be true for this to make sense as stated.”
Presupposition in linguistics is a related cousin — what an utterance requires to be true for it to even be interpretable. The Mimamsa concept is broader: presupposition is a property of utterances; arthāpatti is a method of arriving at unobserved facts about the world from the coherence requirement of any observation.
Where it Shows Up
Debugging AI output. An LLM produces a response that does not cohere with the prompt as written. What hidden context must the model have access to, or be inferring, for this output to make sense? The route is to the unstated assumption — often a system prompt, a memory, or an inference the prompter did not realise was in play.
Reading institutional decisions. Leadership announces a decision with a public rationale that does not hold together. What must be true, and unstated, for this decision to be sensible? The unstated is then known, with appropriate force, by the necessity of coherence.
System behaviour analysis. A system behaves contrary to spec, and the behaviour is reproducible and patterned. What hidden state must be present for the behaviour to make sense? The hidden state is implicated by the requirement of coherence.
Reading user behaviour. A user does X and then Y, and the two are incompatible given the stated user goal. What unstated goal would make both coherent? The method drives the reframing.
Evaluating model claims. A model is reported to perform at a certain level. The reported performance is inexplicable given the model’s known capabilities. What unstated factor — data leakage, prompt structure, evaluation framing — must be in play for the performance to be coherent?
Diagnostic Question
“What must be true, and unstated, for this observation to be coherent?”
IKS Roots
The Sanskrit term is arthāpatti (अर्थापत्ति), literally “the falling-in of meaning” or “implication of an object.” The Bhatta school of Mimamsa (Kumarila Bhatta, 7th c.) defends it as an independent pramana; the Prabhakara school agrees, with technical differences. Dṛṣṭārthāpatti is the perceptual version, śrutārthāpatti the testimonial version. Nyaya subsumes the move under anumāna, arguing the postulation is a form of inference with an implicit pervasion. The Mimamsakas respond that the cognitive structure is different — no instance from which a pervasion has been generalised is in play; the route is the necessity of coherence itself.
See also N1 (the wider system of pramāṇas) and N2 (vyāpti — the structure of inference, which arthāpatti is distinct from).
Further Reading
Kumarila Bhatta’s Ślokavārttika, the arthāpatti section, with Pārthasārathi Miśra’s commentary. The Karl Potter Encyclopedia of Indian Philosophies volume on Mimamsa for a scholarly entry. Stephen Phillips, Epistemology in Classical India, for an accessible introduction. Wilhelm Halbfass on classical Indian epistemology for the broader frame.