What is arthāpatti?

Compared to the pramāṇa of pratyakṣa (perception) or anumāna (inferential reasoning), arthāpatti (postulation or presumption) has received less attention in contemporary secondary literature. This is unfortunate, since it is broadly recognized as a pramāṇa, whether as a sui generis one, as by Mīmāṃsā and Vedānta, or as a type of anumāna, as by Nyāya. As well, it is an important pramāṇa for the interpretation of Vedic texts and its purported non-reducibility to anumāṇa raises interesting questions about the relationship between epistemology and logic. In this short post I want to raise the question of how we should think about arthāpatti by laying out the interpretive territory through secondary literature. Some earlier posts have taken on the major primary texts, so I will leave that aside here.

Basically, arthāpatti is the kind of reasoning that allows us to reason from

(1) “Fat Devadatta does not eat during the day.”


Fat Devadatta eats at night.


(2) Caitra is alive; I see that Caitra is not at home.


Caitra is living somewhere outside of his home.

The central idea is that there is some kind of tension which arises from a new fact being cognized, against the background of another fact, which cannot be resolved without the postulation of third fact. In example (1) above, the tension is between Devadatta being fat and his not eating during the day. In (2), it is between Caitra’s being alive and Caitra’s not being at home. The tension is not strict logical inconsistency, (p & ~p) but something else–just how to analyze this is something on which Indian thinkers differ.

Major Questions and Literature Overview
There are a few questions whose answers will carve up the conceptual territory.

  1. Is there one arthāpatti or many?
  2. If there are many arthāpatti-s, what distinctions are there?
  3. Is arthāpatti (or arthāpatti-s) a pramāṇa? (If many, are all pramāṇa?)
  4. If it is a pramāṇa, is it sui generis or in virtue of reducing to another?
  5. If it is a pramāṇa, what is its “trigger”, doubt or conflict?
  6. Is arthāpatti deductive or non-deductive?

The most familiar distinction in arthāpatti is between śrūta- and dṛṣṭa-, distinguished based on whether the bits of knowledge that jump-start the process are testimonially-given or experientially given. However, one can also distinguish in śrūta-arthāpatti between abhihita and abhidhāna. The former is where what is anupapatti (unintelligible or inexplicable) is the facts expressed and the latter where it is the expression itself.

Kanaujia 1992 rejects the distinction between śrūta and dṛṣta, and instead proposes three major kinds of arthāpatti based on (1) linguistic unintelligibility, (2) factual unintelligibility, (3) unintelligibility of contradiction, subdividing (2) into (a) explanatory unintelligibility and (b) psychological unintelligibility, where there is only a partial clash between two facts. These types do not entirely map onto distinctions in the primary texts. Kanaujia’s aim is to demonstrate that arthāpatti is a heterogeneous concept in the original sources which must be more carefully analyzed. As a result of these distinctions, Kanaujia argues that only some of the sub-types are pramāṇa-s, and some are reducible to anumāṇa.

Bharadwaja 1988 and 1990 (Chapter 4) does not make such a distinction among kinds of arthāpatti but thinks that it is simply a matter of contextual interpretation–and thus not a pramāṇa at all. Chinchore and Chinchore 1984 argue that arthāpatti is a pramāṇa, and in fact a sui generis one (contra the Nyāya position). They, too, emphasize the linguistic nature of arthāpatti, claiming that it is essentially a “semantic gap” which needs to be filled in.

Yoshimizu 2007 says something similar, that arthāpatti is propositional derivation, a kind of reasoning in which one sentence is deduced from another sentence. He proposes a formalization of some of the most common stock examples of arthāpatti, those found in Kumāila, to make his point. Like Yoshimizu, Barlingay 1965 attempts some formalization of arthāpatti, arguing that postulation is a truth-functional deductive argument. He analyzes it in terms of semantic implication, modus ponens, and disjunctive syllogism. Rastogi 1983’s brief treatment (in an appendix to the text) also says something similar, though without the formalization, saying that arthāpatti is a semantic implication whose conclusion is necessary.

One of the most commonly cited and earliest works on arthāpatti is Datta 1932. He defends its independence as a pramāna but unlike Yoshimizu and Barlingay, he thinks that arthāpatti is equivalent to hypothesis and transcendental argument. To the objection that this would strip arthāpatti of its being a pramāṇa (since more than one hypothesis is possible in any given situation), he responds that, just like anumāna can be misapplied, so can arthāpatti. When arthāpatti arrives at the truth, it is a pramāṇa. This is along the lines of Taber’s 2009 review of Yoshimizu’s article, in which he says that arthāpatti is pointing in the direction of non-deductive inference such as inference to the best explanation (IBE). Maiti 1939 says something similar, that there is a difference of degree, but not of kind, between hypothetical reasoning and arthāpatti, due to the greater degree of certainty in the latter. More recently, Kasturirangan et al 2011 have taken up this question, arguing that arthāpatti is not IBE, since its conclusion necessarily follows from the premises, although they agree that it is also not reducible to anumāna.

Moving Forward – Methodological Question
Given this (very brief survey), I want to raise methodological some questions about how to move forward in understanding arthāpatti. Kanaujia says the following (see image):Screen Shot 2015-06-05 at 10.51.03 AM


What should our approach be to understanding arthāpatti in light of its various definitions and examples in the original literature? How should we employ various formal apparatuses in our analysis? Do you have a preferred analysis of arthāpatti that is not represented here?


Barlingay, A Modern Introduction to Indian Logic. New Delhi: National Publishing House, 1965 (reprint 1976). pp.19-21 and 245.

Bharadwaja, V. “The Concept of Arthapatti.” Indian Philosophical Quarterly 15 (2):113 (1988)

Bharadwaja, V. Form and Validity in Indian Logic. Indian Institute of Advanced Study, Simla, in association with Munshiram Manoharlal Publishers, New Delhi, 1990.

Chinchore, Managala and Chinchore, Mangala. “Arthāpatti.” Annals of the Bhandarkar Oriental Research Institute, Vol. 65, No. 1 (1984), pp. 101-113.

Datta, D.M. Six Ways of Knowing: A Critical Study of the Advaita Theory of Knowledge. Calcutta University Press, 2nd Ed. 1960. Book No V, Arthāpatti, p.237.

Kanaujia, Tulsi Ram. “Heterogeneity of Arthapatti.” In Gustav Roth & H. S. Prasad (eds.), Philosophy, Grammar, and Indology: Essays in Honour of Professor Gustav Roth. Delhi: Sri Satguru Publications, 1992. pp.165-184.

Kasturirangan, Rajesh, Guha, Nirmalya and Ram-Prasad, Chakravarthi. “Indian cognitivism and the phenomenology of conceptualization.” Phenomenology and the Cognitive Sciences. June 2011, Volume 10, Issue 2, pp 277-296.

Maiti, M.P. “Arthāpatti and Epistemological Possibilities of Doubt.” The Philosophical Quarterly. Jan 1939. Vol. XIV-IV. p.314-321.

Rastogi, Maharaj Narain. The theories of implication in Indian and Western philosophy : a critical study. Delhi : Bharatiya Vidya Prakashan, 1983.

Taber, John. Review of Karin Preisendanz, ed., Expanding and Merging Horizons: Contributions to South Asian and Cross-Cultural Studies in Commemoration of Wilhelm Halbfass, Wiener Zeitschrift für die Kunde Südasiens, 52-53, 311-315, 2009-10

Yoshimizu, Kiyotaka: “Kumārila’s Propositional derivation without Pervasion” Expansion and Merging Horizons: Halbfass Commemoration Volume, Wien 2007.



About Malcolm Keating

Malcolm Keating is Assistant Professor of Philosophy in the Humanities Division at Yale-NUS College, Singapore.

11 Replies to “What is arthāpatti?”

  1. Malcolm,

    Perhaps at the periphery (if not outside the bounds) of your concerns here, but I’m wondering if you’ve examined how “presumption” is used in legal reasoning, argumentation, and epistemology generally in Western philosophy by way of comparison. Apart from Douglas Walton (and a handful of others, like Nicholas Rescher, who gives it some attention in his epistemological studies), it strikes me that presumption is, so to speak, under-theorized or at least not systematically examined in contemporary philosophy (in comparison, say, to deductive, inductive, and abductive reasoning), however ubiquitous it happens to be in practical reasoning and dialogue contexts. Some years ago I wrote to Ed Zalta as the principal editor of the SEP suggesting someone write an entry on “presumption” but he did not think it was philosophically/analytically worthy of treatment. Any thoughts about this?

    • Patrick, I haven’t looked at this, no. Thanks for the tip. It looks like there might be some interesting connections. For what it’s worth, I have gone back and forth between “postulation” and “presumption” as a translation. Knowing that there is some technical sense to presumption already might mean it is worth avoiding, in fact.

        • True, if one uses “postulation” to translate “kalpanā”, then it isn’t useful for the term “arthāpatti” when the former explains the latter. However, depending on context, I’ve used something like “hypothetical construction” or “intellectual conjecture” for “kalpanā.”

          Theres certainly a balancing act here of finding terms which have the right connotations in English while lacking too heavy theoretical baggage and cohering with other translational choices.

  2. For arthāpatti as an alaṃkāra, Jayadeva says in his Candrāloka:
    arthāpattiḥ svayaṃ sidhyet padārthāntaravarṇanam |
    sa jitas tvanmukhenenduḥ kā vārttā sarasīruhām ||5.37||
    “The a. is the fact that the description of another notion comes by itself. For example : (if already) the moon is vanquished by your face, what about for the lotus ! (it is not necessary to deal with the lotus)”
    Appaya Dīkṣita (KA 120) changes the definition (same example) :
    kaimutyenārthasaṃsiddhiḥ kāvyārthāpattir iṣyate |
    “The establishing of a meaning by the “a fortiori” argument is called the poetical arthāpatti”
    and in his comment he refers to the mīmāṃsaka daṇḍāpūpikā ou daṇḍāpūpa-nyāya (“a method of reasoning in which a self-evident truth is illustrated by saying that a mouse which has eaten a stick is sure to eat a cake” MW) — I would be interested to learn more about the “stick-and-mouse” argument (references ?).
    In this case the French “tombée sous le sens” (as in “cela tombe sous le sens”) is a literal translation which is fitting as well as “présomption”.

    • The maxim of the mouse and the cake is found also in Dharmaśāstra (you can find references in the list of nyāyas in the appendix to Kane). It closely resembles this other nyāya:

      kaimutikanyāya, argument a fortiori (if Z is true, than a fortiori, Y, which is an easier version of Z, must be true). The nyāya is applied, e.g., to the case of whether it would be possible to sentence to death a Brahmaṇa murderer. The solution is that one can and the implication is as follows: “The mention of the word `Brahmaṇa’ [in the passage specifying that Brahmaṇa murderers can be sentenced to death] has the purpose of implying an a fortiori argument. If it is possible to sentence to death even a Brahmaṇa murderer, then even more so in the case of another [murderer] (that is, one who is not a Brahmaṇa)” (brāhmaṇagrahaṇaṃ tu kaimutikanyāyārtham. ātatāyī brāhmaṇo ‘pi vadhyaḥ kimutānya iti mitākṣarāyām, Nārāyaṇa Bhaṭṭa’s Vyavahāramāyukha, chapter 24, ātatāyinaḥ, p. 241 in Kane’s edition).

      • Thank you for the reference to Kane’s useful list (HdhS 5/2, pp. 1339-1351, here p. 1344) and the example (to be found in Vijñāneśvara’s Mitākṣarā ad YājñavalkyaS). So kaimutika-nyāya (or kaimutya A.D.) and daṇḍāpūpikā or daṇḍāpūpa-nyāya would be different names (the latter metaphorical, refering to a popular maxim) for denoting the same a fortiori argument presented as the basis of the arthāpatti figure by several poeticians, starting from Ruyyaka’s AS sūtra 63 :
        daṇḍapūpikayā arthāntarāpatanam arthāpattiḥ /
        In his own (or Mankhuka’s) vṛtti the maxim itself is explained but the term kaimutya (kaimutika, kimuta) is not found ; differently the 13th-14th c. Vidyācakravartin’s commentary to the sūtra starts : yathā daṇḍabhakṣaṇenāpūpabhakṣaṇaṃ kaimutyenāpatati tathārthāntarasyāpatanam arthāpattiḥ /
        I remain interested to found references to the maxim in (early) Mīmāṃsā literature (or anywhere else).

        • I will keep an eye on it. By the way, I am by no means an expert in Alaṅkāraśāstra, but isn’t it sort of fun that arthāpatti describes two very different phenomena?

          • Elisa, I wonder just how very different these phenomena are–after all, it seems that the use of the term in Alaṅkāraśāstra like the other figures named after various pramāṇa, picks out some perceived commonality (such as being based in argument a fortiori). At least that is the interpretation of Gerow in his comments on this in figure as used by Appaya Dikṣita who seems to have introduced the term (I don’t have the citation at hand since my books are all packed but it is in his “A Glossary of Indian Figures of Speech.”

          • And if Devadatta was a mouse…
            I agree with Malcolm that there should be something to find in common between the “post-Mammaṭa” (Gerow p. 333; but the “encyclopedic” authors and their commentators I quoted are pre-Appaya Dikṣita) alaṅkāra and the pramāṇa (the former must have been built on the ground of the latter). Ruyyaka (1st half of the 12 c.?) in the vṛtti distincts arthāpatti from anumāna by the fact that in the former there is no invariable concomitance (vyāpti) between the two things involved: a certain thing being admitted, another automatically follows for the very same reason (samāna-nyāya); and he gives two varieties of arthāpatti : (1) the surmise of a non-contextual from the contextual and (2) the surmise of a contextual from the non-contextual.
            Maybe also the daṇḍāpūpa-nyāya is originally not exactly the kaimutika-nyāya, I mean a pure a fortiori argument only : if the mouse has eaten the stick, “I presume”/”obviously” he has eaten the cake (on the stick).

  3. Thank you, Christophe, for this hypothesis. In a comment on another blog (here), Malcolm pointed out to a further connection between arthāpatti and the world of alaṅkāraśāstra. Worth further investigations…

Leave a Reply

Your email address will not be published. Required fields are marked *