Chapter XXII - Inference

The Buddhist philosophers of the school of Dignāga admit only two pramāṇas (sources or media of valid knowledge), viz., perception and inference. Of these two we have already dealt with the nature, scope and function of perception and we propose to treat of inference in this chapter. Inference has been divided by Dignāga and Dharmakīrti and their followers into two kinds, viz., (1) Inference for the sake of one’s own self and (2) for the sake of others. The former is defined as the deduction of the probandum, i.e., the inferrable thesis, from the reason endowed with threefold characteristics, which are as follows:—

1. The reason or the middle term must abide in the subject or the minor term.
2. The reason must abide only in cases which are homologous.
3. The reason must not abide in cases which are heterologous.

The inferrable thesis is the subject, of which the probandum is sought to be predicated. And the homologue is one which invariably possesses the probandum along with the probans or reason. The heterologue is one which is the opposite of the homologue, that is to say, which does not possess the probandum and the probans at any time.

Now, if a particular reason or probans (hetu) is understood to possess these threefold characteristics, it will invariably and unmistakably lead to the knowledge of the probandum. But what is the means of knowing that a particular probans would possess the triple characteristics?

Dharmakīrti says that only the following ‘reasons’ (hetu) would satisfy this triple condition:—

1. a reason which is identical in essence with the probandum (svabhāvahetu);
2. which is an effect or product; and
3. thirdly non-perception.

In other words, in cases of coexistence, only that probans is the ground of valid inference, which bears a relation of essential identity to the probandum and in cases of succession, the reason must be shown to stand in the relation of causality to the probandum, i.e., as the effect of it. Non-perception (anupalabdhi) is of service only in negative inference, that is to say, when the non-existence of a thing or a quality is to be inferred. But non-perception can have logical value only if all the requisite conditions of perception are present in full except the object, or in other words, when the object is amenable to perception and all other conditions of perception are present intact. Well, but why should these three alone be specified as the ground of valid inference and not any other? The reason is that a thing can prove another thing only if there is a natural relation between the two, in other words, if the relation is unconditional and invariable. And what sorts of relation can be invariable and unconditional? The answer is, that only the relations of causality and fundamental identity can be invariable and unconditional. The Naiyāyikas think that invariable relation is understood from observation of concomitance in presence and absence. But mere concomitance, though observed in hundreds of cases, is not sufficient warrant for its invariability unless and until the relation between the probans and the probandum can be clearly reduced either to causality or essential identity.^[1] We defer the discussion of this all-important problem for the present, as it requires fuller treatment, to which a separate section will be devoted.

The triple characteristic of a valid reason is the conditio sine qua non of true inference and a breach of these conditions severally and jointly will give rise to various cases of fallacies.^[2] But there are fallacies of thesis (paksābhāsa) also. It is therefore relevant to enquire into the nature of the thesis (pakṣa), or the conclusion in the language of Aristotelian logic. The thesis or conclusion has been defined as ‘a (proposition) which is desired by the arguer himself to be established only as the probandum and which is not contradicted by other evidence.’ The clause ‘only as the probandum’ excludes the case of unproven reason, as, e.g., in the syllogism, ‘word is non-eternal, because it is visible,’ the reason ‘visibility’ is unproven, as it does not belong to the subject and so has got to be proved. But though unproven like the probandum (non-eternality), it is advanced not as a probandum alone but as a probans also. The clause ‘by the arguer himself’ is intended to exclude all other possible facts save and except what is intended by the arguer himself. Various other facts may be predicable of the subject, but that alone should be regarded as the thesis which is intended by the arguer himself. The clause ‘which is desired’ is significant. It shows that the thesis is not what is expressed in so many words alone, but even what is implied should also be accepted as the intended thesis. Thus, for instance, when the Sāṃkhya argues, “The eye and the rest are for the sake of another, as they are compounds, like beds and cushions and the like,” it is not expressly stated that the phrase ‘for the sake of another’ means ‘for the sake of the self’ (ātmārtha). But though not expressly stated, that is the implied subject and so should be understood as such.^[3] If the thesis is contradicted either partially or wholly by any such evidence as perception, inference, conceptual knowledge (pratīti) or one’s own statement, it would constitute a case of unsound thesis.^[4]

 

II

The Subject-matter of Inference or the Thesis or Conclusion.

In the previous section we have seen what according to Dharmakīrti should be properly regarded as a correct thesis or conclusion. But Dharmakīrti does not vouchsafe any information as to the exact character of what should be regarded as the conclusion. Inference proceeds through the machinery of three terms, the subject, the predicate (the probandum) and the probans. There must also exist an invariable and unconditional relation between the probans and the probandum. These are the conditions precedent of all inference. But what is the objective and the proper matter of inference? This problem was attacked by Dignāga in his Pramāṇasamuccaya.

Dignāga argues,

“There are some logicians who think that the object of inference is the predicate (the probandum), which is invariably connected with the probans. Others again think it is the connexion that is inferred, and neither the subject nor the predicate, because they are known from other evidence. But both these views are untenable. If the probans is known to be invariably connected with the probandum the latter is already known at the time the connexion was apprehended. If it is contended that the predicate was not known as related to the subject, well, let then the subject thus qualified be regarded as the probandum. The connexion too cannot be the subject-matter of inference, because it does not contain the two terms, the probans and the probandum, in its fold like the subject. Moreover, the relation is not stated as the probandum in the syllogism either by name or through a case-ending. And the probans is not shown to be connected with the relation. Moreover, the relation is tacitly included in the probandum and need not be stated. So it is neither the subject, nor the predicate, nor again the relation that is inferred; but it is the subject as qualified by tbe predicate, which is inferred on the strength of invariable relation between the probans and the probandum observed elsewhere.”^[5]

Uddyotakara does not subscribe to the position of Dignāga; but his arguments are all sophistical and do not carry conviction.

He denies that there is any relation at all between smoke and fire, because he argues that smoke without fire and fire without smoke are observed and this would be impossible if there was an invariable relation between the two. And even if causal relation is conceded, fire can at best be conceived to be the efficient cause (nimittakāraṇa) of smoke, but an effect is not necessarily bound up with its efficient cause. Nor can there be a locus or subject, say the hill, where the smoke and fire could be inferred to be associated together, because Dignāga does not admit any whole in the shape of the hill. And even if the hill as a locus be perceptible, the hill and the smoke are not necessarily found together. If the perception of the hill be a necessary condition of inference, there can be no inference, when a man sees a column of smoke ascending high up in the sky.^[6] But all these arguments of Uddyotakara serve to deny the possibility of inference and this denial does not affect the position of the Buddhist alone, but also of Uddyotakara himself.

And it goes against the verdict of popular experience. Uddyotakara, too, realises this undesirable consequence of his sophistry and so hastens to restrain himself. He avers that there is no contradiction of ordinary experience and inference based upon experience. But there is no logical necessity that the smoke must be perceived in a locus, say, the bill, when a volume of smoke is perceived at a great distance from the source of its origin, high up in the sky rising in an ascending column; in that case the subject-matter of inference is only smoke with fire predicated of it without any reference to the locus. So the proposition of Dignāga that the object of inference is the hill as qualified by the fire is not universally true, though in the generality of cases, Vācaspati concedes, Dignāga’s contention may hold good.^[7]

Kumārila refers to this view of Uddyotakara and justifies it from the charge of petitio principii. If the thesis is of the form ‘smoke is possessed of fire,’ the probans ‘smoke’ is taken as a part of the thesis, and so there can be no inference owing to the lack of another probans. Kumārila says that there is no such apprehension, as the subject is a particular individual smoke and the probans is the smoke-as-universal.^[8] But though he vindicates the view of Uddyotakara from the charge of logical inconsequence, Kumārila however accepts the position of Dignāga without reservation and carries it to greater logical precision.

Though Kumārila’s conception of the thesis is substantially identical with that of Dignāga, Kumārila’s arguments, however, are more elaborate and cogent and so we do not hesitate to reproduce them even at the risk of some repetition. Neither the predicate ‘fire,’ nor the subject ‘hill’.can be severally or jointly be the legitimate object of inference, as the ‘hill’ is known from perception and the ‘fire’ too is known at the time that the universal relation between fire and smoke was apprehended. So inference would be useless, as it would not add to our stock of knowledge. In an inference, where impermanence is sought to be proved of ‘word’ on the ground of its being ‘a product,’ the syllogism cannot be of the form ‘Impermanence exists, since it is a product.’ Because, ‘being a product’ is not an attribute of ‘impermanence’ (anityatva). Nor can it be of the form ‘word exists, because it is a product’ there being no invariable concomitance between word and being a product, as the fact of being a product exists in an earthen jar but word does not. Equally absurd would be the syllogistic form, ‘word and impermanence exist, because of being a product,’ as the probans ‘being a product’ belongs to word alone and not to both.

The possibility of the subject or the predicate being the thesis is thus ruled out of court. Nor can the thesis be supposed to consist in the necessary connexion between the subject and the predicate. Had it been so, the connexion would have been expressed in the thesis either by a whole word or a genitive case-ending in some such form as ‘the hill’s fire exists’ (parvatasyā’gnir asti), or ‘the connexion between the hill and the fire exists’ (agniparvatasambandho’sti). Nor is there an invariable concomitance between the probans and the ‘connexion’ as probandum. Again, connexion as such cannot be the subject of the conclusion, as it does not possess the probans and the probandum inside itself like the hill, which possesses both smoke and fire. And the connexion as particularised (say of hill and fire) cannot be known before the inference is arrived at, and even then, ‘the possession of smoke’ cannot be an attribute of ‘the connexion’ (pakṣadharma). So this alternative, too, should be dropped down. What then is the subject-matter of inference? The answer is that ‘the hill and fire related as subject and predicate, adjective and substantive, is the thesis and so the idea of relation being implicitly contained in the related whole, as a logical presupposition, does not require to be explicitly formulated, though it is the all-important factor, being previously unknown. The reason is that relation in itself without reference to concrete terms is useless. The object of inference is thus a judgment, a relational whole with two factors, the subject and the predicate. But which is the subject and which is the predicate? There are some thinkers who hold that the relation of subject and predicate is one of substantive and adjective and is interchangeable according to option, though the fact is undeniable that one element in a subject leads to the inference of another element. Thus, in the inference of impermanence of word on the ground of its being a product, the thesis can be expressed either as ‘word is impermanent’ or as ‘there is impermanence in word.’ Similarly in the familiar instance of hill and fire, the thesis can assume either form: ‘The fire is in the mountain’ or ‘the mountain has fire,’ the subject and the predicate being left indeterminate concepts and their relation as reversible.

Let us take a concrete syllogism and see the position of the subject and the predicate. ‘Word is impermanent, because it is a product, whatever is a product is impermanent, as for instance a jar.’ In this syllogism, ‘word’ is the subject and the substantive element. Let the syllogism again be as follows:—

 

‘Impermanence is in-word, because it is a product, etc.’

In the latter syllogism, as the fact of being a product cannot be construed with ‘impermanence,’ it has to be construed with ‘word’ though it is a part of the adjectival clause, and hence subordinate. But in the first syllogism, there is no difficulty of construction, as ‘word’ is the subject and the probans ‘being a product’ is directly construed with it. It may be urged that the probans (smoke) is invariably connected with the probandum (fire) and so the existence of smoke can prove the existence of fire. In that case how can it be construed with the subject (hill)? But this need not cause any difficulty—the universal proposition shows the invariable concomitance between fire and smoke independently of the subject and though the probandum (fire) is construed as an adjective of the hill, the connexion of the probans and the probandum in and through the hill is not difficult to understand. Moreover, the hill is apprehended prior to fire, and fire is known only by a process of inference and so the hill should be regarded as the subject, with ‘fire’ predicated of it. And though the subject ‘hill’ was known as such, the hill as qualified by fire was not known. Inference, therefore, is not a repetition of previous information; on the contrary it constitutes a distinct advance in knowledge.^[9]

We have seen that Dignāga’s speculations on the nature of the thesis were full) accepted by Kumārila. Later Naiyāyikas, notably Jayantabhaṭṭa, accepted the position as final. Uddyotakara’s animadversion is rather sophistical and we do not know that any later Naiyāyika has accepted his finding. Even Vācaspati Miśra had to admit the cogency of Dignāga’s arguments and only in special circumstances could he find a justification for the extraordinary conclusion of Uddyotakara. Uddyotakara’s debate is inspired more by spite than logical justice and this is proved by the verdict of posterity, notably of Jayanta, who has given unqualified support to Dignāga’s and Kumārila’s position. It is rather curious that Uddyotakara himself accepts the position of Dignāga in his comments on 1. 1. 33. There is absolutely no logical necessity to rebut the finding of Dignāga, which is, on the contrary, in close conformity with the position of Vātsyāyana. Dignāga only drew out the logical implications of Vātsyayana’s view.^[10] Vācaspati skips over this portion of the Vārttika, perhaps in full consciousness of the palpable inconsistency, which he could not expose as a commentator.

Footnotes and references:

[1]:

kāryakāraṇabhāvād vā svabhāvād vā niyāmakāt |
avinābbāvaniyamo’darśanān na na darśanāt ||
      Pramāṇavārttika, quoted in Tāt. ti. P., 158.

[2]:

Vide History of Indian Logic, under Dignāga and Dharmakīrti.

[3]:

N. B., Ch. III, pp. 110-11.

Cf. anukto’pi vādinā ya eve’cchayā viṣayīkṛtaḥ sa evā’yam sādhya iṣyate.
      T. S. P., p. 672.

[4]:

N. B., Ch. III. For an exposition in English, vide H. I. L, pp. 312-13.

[5]:

kecid dharmāntaraṃ meyaṃ liṅgasyā’vyabhicārataḥ |
sambandhaṃ kecid icchanti siddhatvād dharma-dharmiṇoḥ ||
liṅgaṃ dharme prasiddhāṃ cet kim anyat tena mīyate |
atha dharmiṇi tasyai’va kimarthaṃ nā’numeyatā ||
sambandhe’pi dvayaṃ nāsti ṣaṣṭhī śrūyota tadvati |
avācyo’nugṛhītatvān na cā’sau liṅgasaṅgataḥ ||
liṅgasyā’vyabhicāras tu dharmeṇā’nyatra dṛśyate |
tatra prasiddhaṃ tad yuktaṃ dharmiṇaṃ gamayiṣyati ||
      Tāt. ṭi., p. 180.

[6]:

Vide. N. V., pp. 50-51 and Tāt, ṭi., pp. 180-81.

[7]:

yadi dhūmenā’gner(?) anumānaṃ na bhavet, nanu loko virudhyata iti cet, nāsti virodho dhūmaviśeṣeṇā’gniviśeṣaṇasya dhūmasya pratipādyatvāt...... anumeyo’gnimān ayaṃ dhūma iti.
      Ibid.

Cf. yatra tāvat parvatanitambavartinī dhūmalekbā satatam udgacchantī dṛśyate tatrā’sāv eva tadviśiṣṭo’numīyata iti lokaprasiddham eve’ti kim atra vaktavyam. yatra tu bhūyiṣṭhātayā tasya dhūmasya dūratvena deśo na lakṣyate, dhūma eva tv abhraṃlibo limpann ivā’bhramaṇḍalam avalokyate, tatra deśānumānaprayāsālasatayā dṛśyamāno dhūmaviśeṣa evā’gnimattayā sādhyate.
      Tāt. ṭi., p. 182.

[8]:

prameyatā...... dhūmasyā’nyaiś ca kalpitā | (śl. 47)......
nanu dhūmaviśeṣyatve hetoḥ pakṣaikadeśatā |
nai’tad asti, viśeṣe hi sādhye sāmānyahetutā | śl. 50½-51½
      S. V., anumānapariccheda.

[9]:

S. V., śls 28-48.

deśasya parvatādes tu svarūpe pāvakād ṛte |
gṛhīte’gniviśiṣṭasya punar jñānaṃ na duṣyati ||
tasmād dharmaviśiṣṭasya dharmiṇaḥ syāt prameyatā |
sa deśasyā’gniyuktasya.
      Ibid, 46-48.

[10]:

sādhyanirdeśaḥ pratijñā 1. 1. 33. prajñāpanīyena dharmeṇa dharmiṇo viśiṣṭasya parigrahavacanaṃ pratijñā. (Bhāṣya)—na brūmo dharmimātraṃ sādhyam api tu prajñāpanīyadharmaviśiṣṭo dharmī ’ti.
      N. V., p. 108, et seq.