A History of Indian Philosophy Volume 3

by Surendranath Dasgupta | 1940 | 232,512 words | ISBN-13: 9788120804081

This page describes the philosophy of venkatanatha’s treatment of inference: a concept having historical value dating from ancient India. This is the eleventh part in the series called the “philosophy of the ramanuja school of thought”, originally composed by Surendranath Dasgupta in the early 20th century.

Part 11 - Veṅkaṭanātha’s treatment of Inference

Inference according to the Rāmānuja school is very much the same as inference according to the Naiyāyikas. Inference is the direct result of parāmarśa, or knowledge of the existence of reason (associated with the knowledge of its unblemished and full concomitance with the probandum) in the object denoted by the minor term[1]. Inference is a process by which, from a universal proposition which includes within it all the particular cases, we can make an affirmation regarding a particular case.[2] Inference must therefore be always limited to those cases in which the general proposition has been enunciated on the basis of experience derived from sensible objects and not to the affirmation of ultra-sensual objects—a reason which precludes Rāmānuja and his followers from inferring the existence of Īśvara (God), who is admitted to be ultra-sensual (atīndriya) (ata eva ca vayam atyantā-tīndriya-vastv-anumānaṃ necchāmaḥ)[3].

As formulated by the traditional view of the school, the principle of concomitance (vyāpti) holds that what in the range of time or space is either equal or less than another is called the “pervaded” (vyāpya) or the hetu, while that which in the range of time or space is either equal or greater than it, is called vyāpaka or the probandum[4]. But this view does not cover all cases of valid concomitance. The example given for spatial and temporal co-existence is that between date-juice (rasa) and sweetness (guḍa), or between the shadow thrown by our bodies and the specific position of the Sun. But such spatio-temporal co-existences do not exhaust all cases, as, for example, the sunset and the surging of the sea. This led the later Rāmānujas to adopt a stricter definition of concomitance as unconditional and invariable association (nirupā-dhikatayā niyataḥ sambandho vyāptiḥ)[5].

Regarding the formation of this inductive generalization or concomitance, we find in Tattva-ratnākara, an older authority, that a single observation of concomitance leading to a belief is sufficient to establish a general proposition[6]. But Veṅkaṭanātha urges that this cannot be so and that a wide experience of concomitance is indispensable for the affirmation of a general proposition of concomitance.

One of the important points in which Rāmānuja logic differs from the Nyāya logic is the refusal on the part of the former to accept kevala-vyatireki (impossible-positive) forms of inference, which are admitted by the latter. Thus, in the kevala-vyatireki forms of inference (e.g. earth is different from other elements on account of its possession of smell) it is argued by the Nyāya logic that this difference of earth with other elements, by virtue of its possession of the specific property of smell not possessed by any other element, cannot be proved by a reference to any proposition which embodies the principle of agreement in presence anvaya. This view apparently seems to have got the support of the earlier Rāmānuja logicians such as Varadaviṣṇu Miśra and Bhattārakaguru (in his Tattva-ratnākara); but both Veṅkaṭanātha (in his Nyāya-pariśuddhi) and the author of the Rāmānuja-siddhānta-saṃgraha point out that, since Yāmuna rejects the kevala-vyatireki form of argument in his lecture on Ātma-siddhi, it is better to suppose that, when the previous authors referred to spoke of kevala-vyatireki as a form of inference, it was not admission of their acceptance of it, but only that they counted it as being accepted by the Nyāya logicians[7].

The author of the Rāmānuja-siddhānta-saṃgraha points out that it may very well be brought under anvaya-vyatireki. Thus we may argue “body is earthly by virtue of its possession of smell; for whatever possesses smell is earthly and whatever does not possess smell is not earthly.” So in this form it may be put forward as a anvaya-vyatireki form of argument. The possession of smell (gandhavattva) may very well be put forth as “reason” or hetu, the presence of which determines earthiness and the absence of which determines non-earthiness or difference from non-earthiness.

Rāmānuja logic admits the necessity of “tarka” (cogitation regarding the relative possibilities of the alternative conclusions by a dialectic of contradictions) as an indispensable means of inferential conclusions. Regarding the number of propositions, Veṅkaṭanātha says that there is no necessity of admitting the indispensable character of five propositions. Thus it must depend on the way in which the inference is made as to how many propositions (avayava) are to be admitted. It may be that two, three, four or five propositions are deemed necessary at the time of making an inference. We find it said in the Tattva-ratnākara also that, though five propositions would make a complete statement, yet there is no hard and fast rule (aniyama) regarding the number of propositions necessary for inference[8].

Veṅkaṭanātha urges that inference is always limited to perceptible objects. Things which entirely transcend the senses cannot be known by inference. Inference, though irrefragably connected with perception, cannot, on that account, be regarded as a mode of perception; for the knowledge derived from perception is always indirect (a-parokṣa). Inference cannot also be regarded as due to memory; for it always reveals new knowledge. Further, it cannot be said to be a form of mental intuition, on account of the fact that inference works by rousing the subconscious impressions of the mind; for such impressions are also found to be active in perception, and on that analogy even perception may be called mental intuition.

Vyāpti (concomitance) may be defined as that in which the area of the probandum (sādhya) is not spatially or temporally less than (i a-nyūna-deśa-kāla-vṛtti) that of the reason, hetu — and reason is defined as that, the area of which is never wider than that of the probandum (a-nadhik-deśa-kāla-niyataṃ vyāpyam). As an illustration of spatial and temporal co-existence (yaugapadya) Veṅkaṭanātha gives the instance of sugar and sweetness. As an illustration of temporal co-existence (yaugapadya) he gives the example of the measure of the shadow and the position of the sun. As a case of purely spatial co-existence he gives the instance of heat and its effects. Sometimes, however, there is concomitance between entities which are separate in space and time, as in the case of tides and their relation to the sun and the moon[9].

Such a concomitance, however, between the probandum and the reason can be grasped only by the observation of numerous instances (hhūyo-darśana-gamya), and not by a single instance, as in the case of Śaṅkara Vedānta as expounded by Dharmārājā-dhvarindra. Bhattārakaguru, in his Tattva-ratnākara, in explaining the process by which the notion of concomitance is arrived at, says that, when in numerous instances the concomitance between the probandum and the reason is observed, the result of such observation accumulates as subconscious impressions in favour of the universal concomitance between all cases of probandum and all cases of the reason, and then in the last instance the perception of the concomitance rouses in the mind the notion of the concomitance of all probandum and all reason through the help of the roused subconscious impressions previously formed.

Veṅkaṭanātha admits concomitance through joint method of Agreement and Difference (anvyaya-vyatireki) and by pure Agreement (kevalā-nvayi), where negative instances  are not available. Ordinarily the method of difference contributes to the notion of concomitance by demonstrating that each and every instance in which the probandum does not occur is also an instance in which the reason does not occur. But in the case of kevalā-nvayi concomitance, in which negative instances are not available, the non-existence of the reason in the negative instance cannot be shown. But in such cases the very non-existence of negative instances is itself sufficient to contribute to the notion of kevalā-nvayi concomitance. The validity of kevalā-nvayi concomitance is made patent by the fact that, if the reason remains unchanged, the assumption of a contrary probandum is self-contradictory (vyāhata-sādhya-viparyayāt), and this distinguishes it from the forms of kevalā-nvayi arguments employed by Kulārka in formulating his Mahā-vidyā doctrines.

Rāmānuja’s own intention regarding the types of inference that may be admitted seems to be uncertain, as he has never definitely given any opinion on the subject. His intention, therefore, is diversely interpreted by the thinkers of his school. Thus, Meghanādāri gives a threefold classification of inference:

  1. of the cause from the effect (kāraṇā-numāna);
  2. of the effect from the cause (kāryā-numāna) ;
  3. and inference by mental association (anu-bhavā-numāna —as the inference of the rise of the constellation of Rohiṇī from the Kṛttikā constellation).

As an alternative classification he gives

  1. the joint method of agreement and difference (anvaya-vyatireki);
  2. inference through universal agreement in which no negative instances are found (kevalā-nvayi);
  3. and inference through exclusion, in which no positive instances are found (kevala-vyatireki).

Bhattārakaguru and Varadaviṣṇu Miśra, who preceded Veṅkaṭanātha in working out a consistent system of Rāmānuja logic, seem also to admit the three kinds of inference, viz. anvyayi, kevalā-nvayi, and kevala-vyatireki, as is evident from the quotation of their works Tattva-ratnākara and Māna-yāthātmya-nirṇaya. Veṅkaṭanātha, however, tries to explain them away and takes great pains to refute the kevala-vyatireki form of argument[10]. His contention is that there can be no inference through mere negative concomitance, which can never legitimately lead to the affirmation of any positive character when there is no positive proposition purporting the affirmation of any character. If any such positive proposition be regarded as implied in the negative proposition, then also the contention that there can be inference from purely negative proposition fails.

One of the conditions of validity of inference is that the hetu or reason must exist in the sa-pahṣa (that is, in all such instances where there is the sādhya), but in the vyatireki form of inference, where there are no positive instances of the existence of the hetu and the sādhya excepting the point at issue, the above condition necessarily fails[11]. The opponent might say that on the same analogy the kevalā-nvayi form of argument may also be denied; for there negative instances are found (e.g. idaṃ vācyam prameyatvāt). The reply would be that the validity of a kevalā-nzaxi form of argument is attested by the fact that the assumption of a contrary conclusion would be self-contradictory.

If the contention of the opponent is that the universal concomitance of the negation of the hetu with the negation of the sādhya implies the absolute coincidence of the hetu and the sādhya, then the absolute coincidence of the hetu and the sādhya would imply the absolute coincidence of the opposites of them both. This would imply that from the absolute coincidence of the hetu and the sādhya in a kevalā-nvayi form of inference the absolute coincidence of their opposites would be demonstrable. This is absurd[12]. Thus, the Naivāikas, who admit the kevalā-nvayi inference, cannot indulge in such ways of support in establishing the validity of the kevala-vyatireki form of argument. Again, following the same method, one might as well argue that a jug is self-revealing (sva-prakāśa) because it is a jug (ghatat-vāt) ; for the negation of self-revealing character (a-sva-prakāśatza) is found in the negation of jug, viz. the cloth, which is impossible (van naivaṃ tan naivaṃ yathā paṭaḥ). Thus, merely from the concomitance of two negations it is not possible to affirm the concomitance of their opposites.

Again, in the above instance— anubhūtir ananubhāvyā anubhūtitvāt (immediate intuition cannot be an object of awareness, because it is immediate intuition)—even the existence of an-anubhāvyatva (not being an object of awareness) is doubtful; for it is not known to exist anywhere else than in the instance under discussion, and therefore, from the mere case of concomitance of the negation of an-anubhāvyatva with the negation of anubhūti the affirmation of an-anubhāvyatva would be inadmissible. Moreover, when one says that that which is an object of awareness (anubhāvya) is not immediate intuition, the mere affirmation of the negative relation makes anubhūti an object of awareness in a negative relation, which contradicts the conclusion that anubhūti is not an object of awareness.

If, again, the character that is intended to be inferred by the vyatireki anumāna is already known to exist in the pakṣa, then there is no need of inference. If it is known to exist elsewhere, then, since there is a sa-pakṣa[13], there is no kevala-vyatireki inference. Even if, through the concomitance of the negation of the hetu and the sādhya, the sādhya is known to exist elsewhere outside the negation of the hetu, its presence in the case under consideration would not be demonstrated.

Again, in the instance under discussion, if, from the concomitance of the negation of not being an object of awareness and the negation of immediate intuition, it is argued that the character as not being an object of awareness (a-vedyatva) must be present somewhere, then such conclusion would be self-contradictory; for, if it is known that there is an entity which is not an object of awareness, then by that very fact it becomes an object of awareness.

If an existent entity is ruled out from all possible spheres excepting one, it necessarily belongs to that residual sphere. So it may be said that “willing, being an existent quality, is known to be absent from all spheres excepting the self; it, therefore, necessarily belongs thereto.” On such an interpretation also there is no necessity of vyatireki anumāna ; for it is really a case of agreement (anvayā); and it is possible for us to enunciate it in a general formula of agreement such as “an existent entity, which is absent from all other spheres excepting one must necessarily belong to that residual sphere.” Again, in such an instance as “all-knowingness (sarva-vittva), being absent in all known spheres, must be present somewhere, as we have a notion of it, and therefore there must be an entity to which it belongs, and such an entity is God,” we have the well known ontological argument which is of vyatireki type. Against such an inference it may well be contended with justice that the notion of a hare’s horn, which is absent in all known spheres, must necessarily belong to an unperceived entity which is obviously false.

It may be contended that, if the vyatireki inference is not admitted, then that amounts to a denial of all defining characters; for a defining character is that which is absent everywhere except in the object under definition, and thus definition is the very nature of vyatireki inference. The obvious reply to this is that definition proceeds from the perception of special characteristics which are enunciated as the defining characteristics of a particular object, and it has therefore nothing to do with vyatireki inference[14]. It may also be urged that defining characteristics may also be gathered by joint method of agreement and difference, and not by a vyatireki inference as suggested by the opponents. In such an instance as where knowability is defined as that which is capable of being known, no negative instances are known but it still remains a definition. The definition of definition is that the special characteristic is existent only in the object under definition and nowhere else (a-sādhāraṇa-vyāpako dharmo lakṣaṇaṃ)[15].

In the case where a class of objects is defined the defining class-character would be that which should exist in all individuals of that class, and should be absent in all other individuals of other classes. But when an individual which stands alone (such as God) is defined, then we have no class-character, but only unique character which belongs to that individual only and not to a class. Even in such cases, such a defining character differentiates that entity from other entities (Brahmā, Śiva, etc.) with which, through partial similarity, He might be confused. Thus, the definition is a case of agreement of a character in an entity, and not a negation, as contended by those who confuse it with vyatireki inference. Therefore, the kevala-vyatireki form of inference cannot be supported by any argument.

On the subject of propositions (avayava) Veṅkaṭanātha holds that there is no reason why there should be five propositions for all inference. The dispute, therefore, among various logicians regarding the number of propositions that can be admitted in an inference is meaningless; for just so many propositions need be admitted for an inference as are sufficient to make the inference appeal to the person for whom it may be intended. Thus, there may be three, four, or five propositions, according to the context in which the inference appears.

In addition to inference Veṅkaṭanātha also admits śabda, or scriptural testimony. No elaboration need be made here regarding the śabda-pramāṇa, as the treatment of the subject is more or less the same as is found in other systems of philosophy. It may be remembered that on the subject of the interpretation of words and sentences the Naiyāikas held that each single element of a sentence, such as simple words or roots, had its own separate or specific sense. These senses suffer a modification through a process of addition of meaning through the suffixes of another case-relation. Viewed from this light, the simple constituents of sentences are atomic, and gradually go through a process of aggregation through their association with suffixes until they grow into a total meaning of the sentence. This is called the abhihitā-nvaya-vāda. The opposite view is that of anvitā-bhidhāna-vāda, such as that of Mīmāṃsaka, which held that no sentence could be analysed into purely simple entities of meaning, unassociated with one another, which could go gradually by a process of aggregation or association. Into however simple a stage each sentence might be capable of being analysed, the very simplest part of it would always imply a general association with some kind of a verb or full meaning. The function of the suffixes and case-relations, consists only in applying restrictions and limitations to this general connectedness of meaning which every word carries with itself.

Veṅkaṭanātha holds this anvitā-bhidhāna-vāda against the abhihitā-nvaya-vāda on the ground that the latter involves the unnecessary assumption of separate specific powers for associating the meaning of the simplest word-elements with their suffixes, or between the suffixed words among themselves and their mutual connectedness for conveying the meaning of a sentence[16]. The acceptance of anvitā-bhidhāna was conducive to the philosophy of Rāmānuja, as it established the all-connectedness of meaning (viśiṣṭā-rtha).

Rāmānuja himself did not write any work propounding his views of logic consistent with his system of philosophy. But Nāthamuni had written a work called Nyāya-tattva, in which he criticized the views of Gotama’s logic and revised it in accordance with the Viśiṣṭā-dvaita tradition. Viṣṇucitta wrote his Saṅgati-mālā and Prameya-saṃgraha, following the same lines, Bhattārakaguru wrote his Tattva-ratnākara, and Varadaviṣṇu Miśra also wrote his Prajñā-paritrāṇa and Māna-yāthātmya-nirṇaya, working out the views of Viśiṣṭā-dvaita logic. Veṅkaṭanātha based his Nyāya-pariśuddhi on these works, sometimes elucidating their views and sometimes differing from them in certain details. But, on the whole, he drew his views on the Viśiṣṭā-dvaita logic from the above writers. His originality, therefore, in this field is very limited. Meghanādāri, however, seems to differ very largely from Veṅkaṭanātha in admitting Upamāna and arthāpatti as separate pramāṇas. He has also made some very illuminating contributions in his treatment of perception, and in his treatment of inference he has wholly differed from Veṅkaṭanātha in admitting vyatireki anumāna.

Meghanādāri admits upamāna as a separate pramāṇa. With him upamāna is the pramāṇa through which it is possible to have the knowledge of similarity of a perceived object with an unperceived one, when there was previously a knowledge of the similarity of the latter with the former. Thus, when a man has the knowledge that the cow which he perceives is similar to a bison, and when later on, roaming in the forest, he observes a bison, he at once notes that the cow which he does not perceive now is similar to a bison which he perceives. This knowledge, Meghanādāri contends, cannot be due to perception, because the cow is not before the perceiver; it also cannot be due to memory, since the knowledge of similarity dawns before the reproduction of the cow in the mind. Meghanādāri holds that no separate pramāṇa need be admitted for the notion of difference; for the knowledge of difference is but a negation of similarity. This interpretation of upamāna is, however, different from that given in Nyāya, where it is interpreted to mean the association of a word with its object on the basis of similarity, e.g. that animal is called a bison which is similar to a cow. Here, on the basis of similarity, the word “bison” is associated with that animal.

Meghanādāri tries to explain this by the function of recognition, and repudiates its claim to be regarded as a separate pramāṇa[17]. He also admits arthāpatti as a separate pramāṇa. Arthāpatti is generally translated as “implication,” where a certain hypothesis, without the assumption of which an obscured fact of experience becomes inexplicable, is urged before the mind by the demand for an explanation of the observed fact of experience. Thus, when one knows from an independent source that Devadatta is living, though not found at his house, a natural hypothesis is urged before the mind that he must be staying outside the house; for otherwise either the present observation of his non-existence at his house is false or the previous knowledge that he is living is false. That he is living and that he is nonexistent at his house can only be explained by the supposition that he is existing somewhere outside the house.

This cannot be regarded as a case of inference of the form that “since somewhere-existing Devadatta is non-existent at his house, he must be existent somewhere else; for all somewhere-existing entities which are nonexistent at a place must be existent elsewhere like myself.” Such an inference is meaningless; for the non-existence of an existing entity in one place is but the other name of its existing elsewhere. Therefore, the non-existence of an existing entity in one place should not be made a reason for arriving at a conclusion (its existence elsewhere) which is not different from itself. Arthāpatti is thus to be admitted as a separate pramāṇa.

Footnotes and references:


parāmarśa-janyā pramitir anumitiḥ.


parāmarśa means vyāpti-viśiṣṭa-pakṣa-dharmatā-jñānarn sarva-viśeṣa-saṃ-grāhi-sāmāñya-vyāpti-dhīr api viśesā-numiti-hetuḥ.
p. 97.




deśataḥ kālato vā'pi samo nyūno'pi vā bhavet
sva-vyāpyo vyāpakastasya samo vā’py adhiko'pi vā.
     Ibid. p. 100.




sambandho’yaṃ sakṛd grāhyaḥ pratīti-sva-rasāt tathā
pratītayo hi sva-rasād dharma-dharmy-avadhīn viduḥ.

The author of the Tattva-ratnākara urges that, since the class-concept (e.g. of dhūma-dhūmatva) is associated with any particular instance (e.g. of smoke), the experience of any concomitance of smoke and fire would mean the comprehension of the concomitance of the class-concept of smoke with the class-concept of fire. So through the experience of any individual and its class-concept as associated with it we are in touch with other individuals included within that class-concept —

sannihita-dhūmādi-vyakti-saṃyuktasya indriyasya tad-āśrita-dhūmatvādiḥ saṃyuktā-śritaḥ, tad-āśrayatvena vyakty-antarāṇi saṃyuktāni, etc.

Nyāya-pariśuddhi, p. 105. (Chowkhamba.)


Nyāya-pariśuddhi and Rāmānuja-siddhānta-saṃgraha.




vyāpti is thus defined by Veṅkaṭanātha—

atre’daṃ tattvaṃ yādṛg-rūpasya yad-deśa-kāla-vartino yasya yādṛg-rūpeṇa yad-deśa-kāla-vartinā yenā’vinā-bhāvaḥ tad idam avinā-bhūtaṃ vyāpyam. tat-pratisambandhi-vyāpakam iti.

Nyāya-pariśuddhi, pp. 101-102.


Veṅkaṭanātha points out that Yāmunācārya, also the accredited teacher of Rāmānuja, did not admit the kevala-vyatireki form of inference in his Siddhi-traya.


The typical forms of vyatireki inference are as follows:

anubhūtir ananubhāvyā anubhūtitvāt, ran naivaṃ tan naivaṃ yathā ghataḥ.
pṛthivi itarebhyo bhidyate gandhavattvāt yan naivaṃ tan naivaṃ yathā jalom.

In the above instance an-anubhāvyatra (non-cognizabilitv) belongs only to immediate intuition. There is thus no sa-pakṣa of anubhūti where an-anubhāryatva was found before.


idaṃ vācyaṃ prameyatvāt (this is definable, because it is knowablc) would, under the supposition, imply that the concomitance of the negation of f ācyatva and prameyatva, viz. avācyatva (indefinable) and aprameyatva (unknowable), would be demonstrable; which is absurd, since no such cases are known.


sa-pakṣa are all instances (outside the instance of the inference under discussion) where the hetu or reason is known to co-exist with the sādhya or probandum.


sajātīya-vijātīya-vyavacchedena lakṣaṇam.
quoted in Nyāya-pariśuddhi, p. 143.


Nyāya-pariśuddhi, p. 145.


abhihitā-nvyaye hi padānāṃ padā-rthe padā-rthānaṃ vākyā-rthe padānāṃ ca tatra iti śakti-traya-kalpanā-gauravaṃ syāt.
p. 369.


See MS. Naya-dyu-maṇi. Chapter on Upamāna.

Like what you read? Consider supporting this website: