Analogical reasoning and postulation

Like postulation (arthāpatti), the pramāṇa or instrument of knowledge known as upamāna, often translated as “analogy,” is both fascinating and underdeveloped in contemporary analysis. There are few stand-alone books focusing just on upamāna, although it is frequently treated along with testimony and perception. I suspect this is because, as with postulation, upamāna is often reduced to another more well-accepted pramāṇa–usually verbal testimony, inference, or perception.

Two books devoted only to upamāna are:

Chattopadhyay, Uma. Dishonoured by Philosophers: Upamana in Indian Epistemology. New Delhi, D.K. Printworld, 2009.

Kumar, Shiv. Upamāna in Indian Philosophy. Delhi: Eastern Book Linkers, 1994.

I have been reading just Kumar’s book, since I am waiting for an inter-library loan of Chattopadhyay to arrive. In doing so, I came across a debate about the relationship between postulation and upamāna. By the way, I am not translating the latter term since I am not sure whether “analogy” quite captures what is going on. The question is whether we can explain upamāna as merely an instance of postulation. The target of this discussion is frequently this famous stock example (version updated a bit for color):

Suppose I live in the city of Singapore and while I have seen cows, I have little experience with other animals. A rancher tells me that the gavaya is an animal that’s like the cow. I venture into the forests of Malaysia one day and happen upon an animal that looks like a cow, but isn’t a cow. I remember what this rancher told me and I realize that the animal in front of me is a gavaya. (Or I think, “That is the animal called ‘gavaya.‘)

It’s important to note that upamāna is implicated in language-learning, although that is not its only role for Indian thinkers. I learn what the denotation of ‘gavaya‘ is through testimony, perception, and memory. While pretty much everyone agrees that I have knowledge at the end of this stock example, not everyone thinks that this knowledge is due to some sui generis source, like (most of) the Nyāya and Mīmāṃsā. For instance, the Advaitan Ānandapūrṇa, in his Nyāyacandrikā (who also agrees it is sui generis) argues against an opponent who thinks upamāna is just postulation:

नोपमानं मानान्तरम् अर्थापततितस्तत्सिद्धे: । यद्यनेन सदृशी सा न स्यात् ततस्तया सदृशो ऽयं नोपलभ्यत सदृशस्योभयनिष्ठत्वात् । उपलभ्यते चायं तया सदृशः ।

Upamāna is not another means of knowledge because it is established that it is postulation. If it were not the case that (the cow) is similar (to the gavaya), then this similarity would not be acquired since the similarity is grounded in both (cow and gavaya). Since this similarity (to the gavaya), there is similarity (in the cow).

In reply, Ānandapūrṇa says:

तदसत् असन्निहितसादृश्यधीः स्मृतिरिति प्रतिज्ञाविरोधात् ।

That is not right, because this contradicts the (opponent’s own) assertion: the thought of similarity which is not present is memory.

In other words, the opponent has earlier said that when I encounter the gavaya in the woods, I remember the earlier testimony that the gavaya is similar to the cow. But here the claim is that I know this similarity through postulation. Only one of these can be correct. Further, Kumar notes that Ānandapūrṇa argues that since the similarity which is remembered is not similarity that is currently present to my senses (when I see the gavaya), it is not able to be established. For most Indian philosophers, the Jaina excluded, smṛti or memory is not a source of knowledge. It is implicated in inference and other pramāṇa, but it is itself not a knowledge-source.

Who might argue that upamāna is postulation? We know at least that Udayana argues for this position in his Kiraṇāvali. As a Naiyāyika, he further thinks that postulation is reducible to inference (anumāna) of a particular sub-variety (it seems that Ānandapūrṇa’s interlocutor does as well, given what Kumar says about other counter-arguments involving a discussion of invariable concommitance or vyāpti). This view is not the standard view of Nyāya, however, as many them prefer to hold that upamāna is sui generis (Jayantabhaṭṭa is another exception, arguing against the Mīmāṃsā that upamāna is a matter of inference).

This discussion is interesting to me, since I find plausible Mukulabhaṭṭa’s arguments that the cognition of (at least some varieties of) figurative language are driven by postulation. When one thinks of figurative language, analogy is central. Poetic thought is about juxtapositions which point at similarities or contrasts, as well as explicit statements of these connections, through similes and the like. Analogical reasoning, too, is important in legal theory, scientific thought, and so on.

In Western philosophical thought, analogy is often characterized something like this:

  1. S is similar to T in at least one respect.
  2. S has another feature, P.
  3. Therefore, T has the feature P, or a feature similar to P in some respect.

An important question, as in Indian philosophical thought, is how to analyze the structure of this argument. Unlike Indian philosophical thought, there is not necessarily a reference to a thinker who is having a mental episode, moving from testimony to perception to memory of the testimony and finally to knowledge of the similarity. Further, the conception of analogy in Western philosophy is inductive: (3) is not a necessary conclusion drawn from (1) and (2) but is probabilistic (or some other model). In contrast, for Indian philosophers, a pramāna results in secure knowledge, not belief to a degree or the like. However, this knowledge need not be secure because it is inferential. After all,  anumāna or inference requires a necessary pervasion relationship (vyāpti) between a sign and what it establishes. This relationship is difficult to characterize adequately for upamāna. Here is how Jayantabhaṭṭa analyzes it in the Nyāyamañjarī:

  1. Sādhya (what is to be established): the cow remembered is similar to the gavaya.
  2. Hetu (inferential sign): the cow has the same kind of limbs as the gavaya. 
  3. Pakṣa (location relating sādhya and hetu): the cow remembered in the forest.

Here, though, in contrast to the Western characterization, what is being discovered is the initial similarity between source S and target T. At least as Jayantabhaṭṭa wants to put it, what comes to be known is that the cow and the gavaya are similar. This poses a problem since it seems as if this was already known through testimony! Further, if the paradigmatic example of a hetu is smoke which indicates fire’s presence, (1) we need to have multiple instances of this in order to make a universal generalization, and (2) for the hetu to be valid, we also need its absence–smoke should not exist where there is not fire. But in this example, I have only seen one instance of a gavaya and there are other animals with limbs like a cow that are not gavaya-s.

There is much more to say about upamāna, postulation, inference, testimony, and their interrelations, as well as the connections between these topics and analogical thought. However, this should give a sense of the topic in brief.

Cross-posted to the my personal blog.

Edited for error: “Jayantabhaṭṭa is another exception, agreeing with the Mīmāṃsā that upamāna is a matter of inference” should read “Jayantabhaṭṭa is another exception, arguing against the Mīmāṃsā that upamāna is a matter of inference.”

About Malcolm Keating

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

8 Replies to “Analogical reasoning and postulation”

  1. Thank you, Malcolm!
    1) I am not sure I understand your point re. the fact that Jayanta, like Mīmāṃsakas, includes upamāna in anumāna. I would have thought that Mīmāṃsakas count upamāna as a separate pramāṇa.
    2) What exactly is known through upamāna is in fact an important point. In order for upamāna to achieve a distinct goal, it cannot be the same similarity which is known through testimony. Kataoka 2003 (JIPh) explains the different positions of Mīmāṃsakas and Naiyāyikas in this regard. Anyway, if the rancher tells you that a gavaya is like a cow, it seems that you must realise through upamāna that the cow resembles the gavaya.
    3) We discussed this already, I think, but let me recommend again Lars Göhler’s book for the hypothesis that upamāna is a Brāhmaṇa-heritage (the book is reviewed here:, see especially p. 168 of the review for upamāna).

    (I slightly corrected the Sanskrit, hope you don’t mind.)

  2. Elisa, thanks for putting in the avagraha — I am not sure how it became an apostrophe. And you’re right, (1) despite introducing the Mīmāṃsā view as that upamāna is sui generis, I contradicted myself later–I can only assume that infelicity is due to sloppy editing on my part. It’s fixed now. They do certainly count upamāna as its own pramāṇa, and Jayantabhaṭṭa is targeting that view in his attempt to provide a syllogism.

    (2) As to the more substantial point about what is known by upamāna, that is crucial, you are right! While the Nyāya and Mīmāṃsā agree that upamāna is sui generis, they disagree about its object. So for instance, Kumārilabhaṭṭa in the ŚV argues that what comes to be known is either the remembered cow as qualified by its similarity to the gavaya or the similarity qualified by the remembered cow. But for Vātsyāyana and Uddyotakara, for instance, what is known is the relationship of the name “gavaya” with its object.

    (3) I don’t remember you mentioning Lars Göhler’s work, but I will certainly take a look.

  3. Michael Reidy has a further post on this topic (here:, which also reminds me of the problem I raised while discussing Göhler’s book, namely that upamāna (in the non-technical sense) appears to have been very relevant in the Brāhmaṇas and it is very relevant in our everyday use. This accounts for its inclusion among pramāṇas, although this inclusion meant that it needed to deliver sure knowledge, and thus devoided it of its ampliative potentiality and restricted it to a few stock examples.

    • Yes, the fact of the stock examples is really interesting. I was reading the Jainatarkabhāṣā today, and even in this 17th century text, no new examples have been adduced, but the discussion (in the context of the Jaina claim that recognition is a pramāṇa) focuses still on the cow example.

      • Adherence to the canonical example can be the result of an uncertain grasp of the principle. One stays close to the shore fearing shipwreck. The adding of details such as the testimony of the forest dweller and even the naming of the beast is perhaps unnecessary. In my understanding of the pramana – ‘like a cow, but not a cow’ is at the heart of it. You get the ‘Bos’ part but not the ‘Gaurus’ part which would come after seeing a herd of them.

        To state the obvious: unless you already have the capacity to discern similarity how could you be taught it? That is a major theme of Philosophical Investigations with its discussion of what makes a game a game, family resemblances, completing series etc. I would add as a correlative the ‘odd one out’ puzzle. This is all commonplace conceptual brachiation.

        • Michael, I agree that the lack of fresh examples indicates lack of grasp of what it really means to make a sound upamāna. I would only add that the first who introduced it in the system must have had an idea of its importance as a euristic tool, in the way you describe, but then in its institutionalisation as a pramāṇa this euristic ability practically went lost in a too strict formalisation (which might be a further reason for the lack of fresk examples).

        • “The adding of details such as the testimony of the forest dweller and even the naming of the beast is perhaps unnecessary.”

          I think that depends, though, upon the view of upamāna? For the Nyāya, it is learning the denotation of “gavaya” which is crucial. I take it that one of their insights is into how we learn language–through some combination of testimony and perception which (on their view) results in something not reducible to either.

          The paradox of inquiry you point to is resolved in different ways by Indian philosophers and is relevant to upamāna, as well, but perhaps that’s the subject of another post.

Leave a Reply

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