Tarka as a cognitive validator

J Indian Philos DOI 10.1007/s10781-011-9148-6

Tarka as Cognitive Validator Nirmalya Guha

Ó Springer Science+Business Media B.V. 2011

Abstract The meaning of the term ‘tarka’ is not clear in the modern literature on Classical Indian Philosophy. This paper will review different modern readings of this term and try to show that what the Nya¯yasu¯tra and its classical commentaries called a ‘tarka’ should be understood as the following: a tarka is a cognitive act that validates a content (of a doubt or a cognition or a speech-act) by demonstrating its logical fitness or invalidates a content by demonstrating its logical unfitness. A tarka can act as a metatheory too. Generating certainty is, according to the Classical Nya¯ya, a job assigned to an epistemic instrument (prama¯n: a). It fails to do so when there arises a doubt regarding it. The moment a tarka dispels the doubt, the epistemic instrument generates certainty. Tarkas of different types will be exemplified by critically ana_ ´a’s applications of tarka in his magnum opus Tattvacinta¯man: i. These lyzing Ganges examples will clarify the definition of tarka formulated in this paper. Keywords

Tarka  Cognitive validator  Metatheory  Old Nya¯ya

For a number of scholars, tarka is something like reductio ad absurdum.1 They understand it in the sense in which the Neo Naiya¯yika understands it. For some scholars, it is ‘counterfactual reasoning’.2 Maybe the second understanding is a little broader than the first one. For Matilal (1986, p. 79; 1998, p. 46), tarka is a reasoning based on a priori principles. Kang (2010) wanted to understand the essence of tarka in terms of three features. I shall try to review these positions and finally suggest a 1 2

See Davis (2001, p. 154). See Mohanty (2000, p. 164).

N. Guha (&) Department of Humanities and Social Sciences, Indian Institute of Technology Kanpur, Kanpur 208016, UP, India e-mail: [email protected]

123

N. Guha

general definition of tarka. In order to support my claim, I shall critically examine _ ´a’s applications of tarka. Ganges For Gotama (su¯tra-ka¯ra), Va¯tsya¯yana (bha¯s: ya-ka¯ra), Uddyotakara (va¯rtika-ka¯ra) and Va¯caspati (ta¯tparya-t:¯ıka¯-ka¯ra), a tarka is a cognitive act or reflection (u¯ha) of some kind that does not generate certainty (aparicchedaka), but strengthens one of the options that constitute a doubt regarding a special feature of an under-informed object. Here is a rough translation of two aphorisms from the Nya¯yasu¯tra of Gotama: Liberation (nih: s´reyasa) is attained through the knowledge of the true nature (tattva) of epistemic instruments (prama¯n: a), the objects represented by valid cognitions (prameya), doubt (sam : s´aya), tarka etc. . .. Tarka is a form of cognitive act or reflection, which operates through ka¯ran: opapatti in order to know the true nature (tattva) of an object that is not known fully.3 The term ‘ka¯ran: opapatti’ has to be carefully translated, and will be translated later in this paper. We may begin our discussion by presenting the following example of tarka originally cited by Va¯tsya¯yana and clarified by Uddyotakara. One may have the following question: Do somebody’s actions regulate their birth? There are two possible answers; ‘yes’ and ‘no’. How does one know which option is correct? One argues in the following way: (1) We see that a certain thing is caused by a specific factor. If we do not accept the universality of causal regulation, then we have to say that anything could be caused by anything else. (2) Everybody’s birth is different from anybody else’s. Somebody is born as a prince and somebody as a pauper! There must be something that causes the difference. (3) The earth, space, time, natural laws etc. are common to all in the sense that everybody is equally entitled to them. So they cannot be the reason for the difference in births. (4) One’s actions or deeds (karman: ) are as unique as one’s birth. X’s deeds and birth exclusively belong to X. (5) Therefore, we can say that one’s action is responsible for one’s birth. Thus the form of the aforementioned tarka should be this: If one’s actions are not responsible for one’s birth, then we cannot account for the fact that everybody’s birth is different from anybody else’s.4 If one argues that something other than somebody’s actions regulates their birth, then one’s argument will have many defects that include profligacy (un-economy), reductio ad absurdum etc. The tarka that has been applied here prefers one option, viz., ‘one’s action is responsible for one’s birth’ to the other. This strengthened (anugr: hı¯ta) option gets validated. The conclusion of the tarka should be true, provided the data supplied to it are correct.

3

prama¯n: a-prameya-sam : s´aya-prayojana-dr: s: :ta¯nta-siddha¯nta-avayava-tarka-nirn: aya-va¯da-jalpa-vitan: d: a¯hetva¯bha¯sa-cchala-ja¯ti-nigrahastha¯na¯na¯m : tattvajña¯na¯n nih: s´reyasa¯dhigamah: . Nya¯yasu¯tra 1.1.1; avijña¯tatattve ‘rthe ka¯ran: opapattitas tattvajña¯na¯rtham u¯has tarkah: . Nya¯yasu¯tra 1.1.40; see Nyaya-Tarkatirtha and Tarkatirtha (1936, 28, 320).

4

katham : punah: karmanimittam : janma? bhedavattva¯t. . . .. so ’yam : bhedah: anekam avasthitam anityam ekadravyam : pratya¯tmaniyatam : nimittam antaren: a na yuktah: . kim : ka¯ran: am? pr: thivya¯dı¯na¯m : sarvapurus: asa¯dha¯ran: ya¯t, pr: thivya¯digatasya ca niyamahetor abha¯va¯t, sarvam : sarva¯tmakam : prasajyeta. dr: s: :tas´ ca niyamabhedah: , tasma¯t karma niya¯makam iti. Nya¯yava¯rtika of Uddyotakara on Nya¯yasu¯tra 1.1.1; Ibid, p. 54.

123

Tarka as Cognitive Validator

Here at this point we encounter two important questions: (1) What does a tarka do? (2) How does it operate? I will try to answer these questions following the Old Nya¯ya tradition. Suppose the cognizing subject S (prama¯tr: ) knows an object O partially and wants to know whether O has a particular property (dharma) P or not. There are two options available to S; either O has P or not. Each option here is a possible epistemic instrument (prama¯n: a) since one of them is valid (pra¯ma¯n: ika). I.e., some epistemic instrument must support it. When there are two contrary options in front of S, a tarka may be applied. The tarka would weigh the two options, and tell S that ‘this option is tenable whereas the other one is not’ (yukta¯yuktavica¯rah: , 5 idam : yuktam idam ayuktam iti). And this is what a passage from Ta¯tparya-t:¯ıka¯ says about the cognitive operation of a tarka: [Sometimes] possible epistemic instruments (prama¯n: a¯ni) that look for the true nature (tattva) [of an object] require an auxiliary factor (itikartavyata¯) that will validate them. A tarka examines the possible epistemic instruments. It validates (anugr: hn: a¯ti) the logically fit instrument and rejects the others. Such a validated epistemic instrument is sufficient for knowing the true nature of an object. . .. One perceptually apprehends a bare floor and the absence of pot. [But still the bare cognition of the absence of pot does not directly lead one to the cognition, ‘there is no pot on this floor’.] One comes to know that there is no pot on the floor through the following tarka: ‘Had there been a pot on this floor, I would have seen it; for it would be as visible as the floor. I cannot see any pot here. Therefore there is no pot here.’6

Epistemic Mechanism in Nya¯ya: A Prerequisite for Understanding Tarka I may briefly state the epistemic mechanism for generating a cognition according to Nya¯ya.7 A valid cognition (prama¯) is a non-accidentally true cognition which the cognizing subject does not doubt.8 It is not a lucky guess. The Naiya¯yika thinks that there are just four epistemic ways in which a valid cognition can be generated. Those four ways are epistemic instruments or prama¯n: as; namely pratyaks: a (perception), anuma¯na (inference), s´abda (testimony) and upama¯na (analogy). Every (ordinary) cognition is an effect (ka¯rya) since it has a beginning. And every effect needs to be produced by a set of causes and conditions (ka¯ran: a-sa¯magrı¯). The 5

yah: prama¯n: a¯na¯m : vis: ayas tam : vibhajate. kah: punar vibha¯gah: ? yukta¯yuktavica¯rah: , idam : yuktam idam ayuktam iti. yat tatra yuktam : bhavati, tad anuja¯na¯ti, na tv avadha¯rayati. anavadha¯ran: a¯t prama¯n: a¯ntaram : na bhavati. Nya¯yava¯rtika on 1.1.1; Ibid, p. 54.

6

etad uktam : bhavati prama¯n: am : tattva¯vadha¯ran: a¯ya prayuktam : karan: ataya¯ itikartavyata¯m apeks: ate. tarkas´ ca prama¯n: avis: ayayukta¯yuktavica¯ra¯tma¯ prama¯n: am : yukte tattve pravartama¯nam anuja¯nan prama¯n: am anugr: hn: a¯ti. tadanugr: hı¯tam : prama¯n: am : tattvanirn: aya¯ya parya¯ptam. . . . tatha¯ hi pratyaks: am eva ta¯vad bhu¯tale pravartama¯nam : yady atra¯bhavis: yad ghat:o : tadvis´es: an: ataya¯ ghat:a¯bha¯ve ’pi pravartama¯nam bhu¯talam iva adraks: ata tena saha tulyadars´anayogyatva¯t, na ca dr: s´yate tasma¯n na¯stı¯ti tena tarken: a anujña¯yama¯nam : ghat:a¯bha¯vavis´is: :tabhu¯tale pravartate kevalam evedam : bhu¯talam : neha ghat:a iti. Ta¯tparya-t:¯ıka¯ of Va¯caspati on 1.1.1; Ibid, p. 54. 7 ‘Cognition’ means ‘jña¯na’ or ‘a piece of cognition’. Let us consider ‘cognition’ a count noun. 8 It has to be a doubt-free cognition or nis´caya.

123

N. Guha

principal cause which actually generates the effect through some causal activity (vya¯pa¯ra) is the karan: a or the causal instrument. In the case of felling a tree, the causal instrument is an axe which through a dynamic connection between itself and the tree finally fells the tree. In the epistemic context, the causal instrument is the prama¯n: a or epistemic instrument. For perceptual cognitions, the instrument is a sense organ such as an eye or an ear. It generates the perception by getting connected (sannikr: s: :ta) with an object. This activity of getting connected is the vya¯pa¯ra or causal activity. In the case of a (hypothetico-deductive) inference (anumiti) having the form, ‘Z has Y, since Z has X’, the instrument is the cognition, ‘anything that has X has Y too’ (which is the vya¯pti-jña¯na or cognition of pervasion). We say that ‘Y pervades X’ if and only if anything that has X has Y too. (In the Indian model) fire pervades smoke since anything that has smoke has fire too. Here fire is the pervader (vya¯paka) and smoke the pervaded (vya¯pya). The cognition of pervasion is the instrument that finally generates the inferential cognition by getting cognitively connected with the Site, i.e., the locus where something is getting inferred. One who knows that anything that has smoke has fire too, sees smoke in a hill and cognizes that ‘this hill has smoke which is pervaded by fire’. This cognition which connects the cognition of pervasion with the Site, i.e., the hill in this case is called a para¯mars´a or ‘consideration’. Immediately after the consideration is born, does one inferentially cognize that ‘this hill has fire’.9 Thus the cognition of pervasion generates the inferential cognition through generating a consideration which is the causal activity (vya¯pa¯ra). In the case of testimonial cognition (s´a¯bda-jña¯na), the instrument is the knowledge of linguistic units (pada-jña¯na) used by the speaker. An epistemic instrument need not always generate a valid cognition (prama¯); one’s eye can take a rope for a snake. That means at least in the Old Nya¯ya tradition, an epistemic instrument (prama¯n: a) is not factive. Tarka and Validation Va¯tsayana says: When an object O is not known fully, the cognizing subject desires to know O as it is. Then there arises the following query that states the two contrary options separately: Does O have the property P, or not? [A tarka] then validates one option through ka¯ran: opapatti. ‘It is logically fit that this option must be supported by an epistemic instrument’—this is ka¯ran: opapatti. Through the demonstration of logical fitness one argues that this option is true while the other one is not.10 9 I am discussing anumiti somewhat elaborately because it will play a crucial role in understanding tarkas. 10

avijña¯yama¯natattve ’rthe jijña¯sa¯ ta¯vaj ja¯yate ja¯nı¯yemam artham iti. atha jijña¯sitasya vastuno vya¯hatau dharmau vibha¯gena vimr: s´ati—kim : svid ittham a¯hosvin nettham iti. vimr: s´yama¯nayor dharmayor ekam : ka¯ran: opapattya¯ anuja¯na¯ti, sambhavati asmin ka¯ran: am : prama¯n: am : hetur iti ka¯ran: opapattya¯ sya¯d evam etan netarad iti. Nya¯yabha¯s: ya of Va¯tsya¯yana on 1.1.40; see Nyaya-Tarkatirtha and Tarkatirtha (1936, pp. 321–322).

123

Tarka as Cognitive Validator

The options that the query consists of mutually block each other since what underlies the query is the doubt: ‘Is this so or not?’ Neither of them generates any certainty. In such a situation, a tarka enters the stage and strengthens one option and dismisses its blocker.11 It provides exactly one option with a causal activity (vya¯pa¯ra) and validates it or dismisses an option.12 It is not a separate epistemic instrument since it cannot apprehend any feature of the object independently; it is just a validator (anugra¯haka).13 In the case of ‘the regulator of birth’, we have seen that a tarka strengthened one option and dismissed the others. In the example cited in Ta¯tparya-t:¯ıka¯, one can see a bare floor; but seeing a bare floor does not immediately inform one that there is no pot on it. One needs to ‘reflect’ a bit in order to have the cognition, ‘there is an absence of pot in this floor’. That reflection is a tarka which has been mentioned in the example. I would like to add another point. We shall consider many examples of tarka in this article. I think that the cases that involve tarkas are ‘unobvious’. The raw perceptions do not depend on tarka. Some (hypothetico-deductive) inferences do. We shall see such inferences. In order to know that ‘this is a floor’ one does not invoke any tarka since this is a matter of basic perception. But that ‘there is no pot here on this floor’ is not as obvious as ‘this is a floor’. The former demands some amount of reflection. The cases that demand some reflection for accepting or dismissing an option may invoke a tarka. A tarka can serve as a metatheory too. A theory can be seen as a network of complex information brought forth by several epistemic instruments. Gotama says: ‘In a discussion that tries to understand the truth about something (tattva), one accepts a thesis or rejects another one on the basis of the given epistemic instruments and tarka. The thesis one accepts should not go against one’s philosophical position (siddha¯nta-aviruddha)’.14 Here Gotama is not just talking about mere empirical data or anything obvious. That means, according to him, a tarka does not only operate at the basic epistemic level, i.e., the level of basic-data-generation, it also operates at the level of theory-building. Va¯da is a discourse in which two sides of a debate try to understand the true nature of something. In a va¯da, one rejects one’s opponent’s theory when there is a tarka that would support one’s position and reject the opponent’s position.

Strong and Weak Tarkas In his Sarvadars´anasam : graha, Ma¯dhava¯ca¯rya says that [according to the Aks: apa¯dadars´ana, i.e., Old Nya¯ya] tarkas are of eleven types.15 Perhaps before the time of 11 viruddhadharmavyuda¯sena avirodhayan ity arthah: . Ta¯tparya-paris´uddhi of Udayana; see Dravida and Dvivedin (1911, p. 329). 12

savya¯pa¯rı¯karoti. Ibid, p. 329. prama¯n: am : paricchedakam : , na tarkah: . tasma¯n na prama¯n: am. . . prama¯n: avis: ayavibha¯ga¯t tu prama¯n: a¯na¯m anugra¯hakah: . Nya¯yava¯rtika on 1.1.1; see Nyaya-Tarkatirtha and Tarkatirtha (1936, pp. 53–54). 13

14

prama¯n: atarkasa¯dhanopa¯lambhah: siddha¯nta¯viruddhah: pañca¯vayavopapannah: paks: apratipaks: aparigraho va¯dah: . Nya¯yasu¯tra 1.2.1; Ibid, p. 335. 15 sa ca eka¯das´avidhah: , vya¯gha¯ta-a¯tma¯s´raya-itaretara¯s´raya-cakraka¯s´raya-anavastha¯-pratibandhikalpana¯-kalpana¯la¯ghava-kalpana¯gaurava-utsarga-apava¯da-vaija¯tya-bheda¯t. See Sharma (1978, p. 467).

123

N. Guha

Udayana, the author of Nya¯ya-va¯rtika-ta¯tparya-paris´uddhi, the notion of ‘tarka’ underwent a process of semantic contraction. At the time of Udayana, it did not any more mean ‘the cognitive act that validates an epistemic instrument’ (prama¯n: a¯nugra¯haka). It started meaning ‘prasan_ga’ which is some sort of a counterfactual imposition (a¯ropa). It is something like the following. Suppose one wants to know whether Y is a fact. There are two ways one can proceed now. (1) Indirect Method: One may begin by assuming the denial of Y. If the denial of Y logically leads one to the denial of X, which is an established fact, one would conclude that Y is a fact too. Here by counterfactually asserting the denial of Y one ‘asserts’ the denial of a fact (here X). Denial of a fact is undesired (anis: :ta) and it has been (counterfactually) imposed (prasakta) on a given case. This is an anis: :ta-prasan_ga or ‘an imposition of the undesired’. (2) Direct Method: One can assume Y itself. If the assumption of Y insists one to contradict X, then Y cannot be true since contradicting an established fact is undesired. Not all the tarkas are based on such impositions. In Ta¯tparya-paris´uddhi, Udayana says that there are tarkas of five types; the vicious circularities of three types (a¯tma¯s´raya, anyonya¯s´raya and cakraka¯s´raya), infinite regress (anavastha¯) and reductio ad absurdum (tadanya-ba¯dhita¯rtha-prasan_ga). In Nya¯ya-su¯tra-vr: tti, Vis´vana¯tha explicitly says that these aforementioned five tarkas are genuine tarkas; the law of parsimony (la¯ghava) etc. are not genuine tarkas since they do not involve any counterfactual imposition (prasan_ga).16 We may briefly see how reductio ad absurdum, vicious circularities and infinite regress involve a counterfactual imposition. Any reductio argument begins by assuming something, say X. Then it shows that if X is true then Y too has to be true. But it is an established fact that Y is not true. Thus assuming X will lead one to assuming Y that is undesired or absurd. Here is an a¯tma¯s´ryaya (self-referential) tarka that uses a direct method: Suppose somebody argues that ‘this pot is caused by itself’ (ayam : ghat:a etad-ghat:a-janyah: ). This statement will be rejected on the basis of the following tarka: If this pot is caused by itself, then let it pre-exist itself, (since a cause has to pre-exist its effect).17 Thus the counterfactual imposition of the property of being self-caused invokes another imposition, namely that of self-pre-existence. The invoked imposition is undesired (anis: :ta) here. Regarding infinite regress, we may remember that the Indian Logician would say that the chicken-egg-problem does not involve any regress since it is valid (pra¯ma¯n: ika). The chicken-egg series is infinite, or at least semi-infinite (in the sense that it may have an end despite not having a beginning), and its infinity does not invoke any undesired factor. But in a theory, not-ending-anywhere can be a problem. Let us consider the following case. Suppose somebody says that in order to be aware of any cognition (jña¯na) C1 one has to have a secondary cognition C2 which is about C1. This view is bad since it involves infinite regress. If in order to be aware of C1, one needs C2, then one needs C3 too in order to be aware of C2 and so on. Thus one never gets aware of C1. But it is evident that we are aware of (at least 16

. . .la¯ghavagaurava¯dikam : tu prasan_ga¯na¯tmakatva¯n na tarkah: . Gautamı¯ya Vr: tti on 1.1.40; see NyayaTarkatirtha and Tarkatirtha (1936, p. 327).

17 This statement is a simplified version of the following: yady ayam : ghat:a etadghat:ajanyah: sya¯t tada¯ etadghat:a¯nadhikaran: aks: an: ottaravarttı¯ na sya¯t. Gautamı¯ya Vr: tti on 1.1.40; Ibid, p. 325.

123

Tarka as Cognitive Validator

most of our) cognitions. The aforementioned view counterfactually imposes that no one is aware of one’s own cognitions which contradicts an established fact. Tarkas that are based on prasan_ga or counterfactual impositions can be called strong tarkas. Tarkas that are not based on counterfactual impositions can be called weak tarkas. The most used strong tarka is the reductio argument. That is why for most of the Neo Naiya¯yikas, the primary meaning of ‘tarka’ has been the reductio argument. Those who think that ‘tarka’ is actually ‘counterfactual reasoning’ have adopted the aforementioned view of Udayana and Vis´vana¯tha, viz., ‘there are tarkas of five kinds’.

Tarka as a Reasoning Based on a priori Principles Matilal (1986, p. 79) writes that tarka ‘‘is rather an argument where we use generally the a priori principles only, or what may be closest to the a priori principles in the Indian tradition.’’ Kang does not think that Matilal’s view on tarka was correct.18 I would like to illustrate the merit of Matilal’s viewpoint. A tarka can be applied to an epistemological content. But it has got an abstract from, a schema. Davis has shown the general schema for tad-anya-ba¯dhita¯rtha-prasan_ga tarka or reductio argument. His schema will be presented in the section on reductio arguments. The following can be an example of vya¯gha¯ta or contradiction: X is both Y and non-Y. The Sanskrit definition of each tarka is schematic. Vis´vana¯tha defines anyonya¯s´raya or mutual dependency as follows: ‘‘tad-apeks: a¯peks: itva-nibandhano ’nis: :ta-prasan_go ’nyonya¯s´rayah: ’’.19 This means the following: ‘X’s dependence on Y, which in turn depends on X, is the undesired feature that characterizes an anyonya¯s´raya’. Such formulations use third-personal pronouns (such as ‘yat’, ‘tat’, etc.) that serve as variables. Epistemological contents can replace the variables in a specific application of tarka. But the tarka-schemata are always a priori. I think this a priority is suggested by the very phrase ‘anis: :ta-a¯patti’. It refers to the undesired factor a tarka brings forth. The phrase suggests that any cognizing subject knows a priori which factors are desired and which are undesired. Contradicting an established fact is undesired; therefore any strong tarka demonstrates that the assumption that has led the arguer to such a contradiction is to be denied. Conceptual profligacy or un-economy (gaurava) is undesired; therefore a la¯ghava tarka or an argument based on economy prefers the most economical option to the 18

Kang (2010, pp. 11–12): Furthermore, what is crucial is the fact that the cognition that initiates the whole procedure of tarka is itself a perceptually based one, albeit one that is insufficiently informative. If this initiating cognition is counted as part of a tarka, one will run into serious problems manoeuvring tarka anywhere near to a priori-based cognition. If we want to interpret the explanation given in the NBh to mean that the tarka is epistemologically void of all experiential cognition, then the only way to do so would be to constrict the procedure so narrowly as to include the logical analysis (vibha¯ga) of the given two alternative possibilities but to exclude any cognitive method for supplying factual data that support either of the two. But in any case this does not keep the basic data underlying the tarka from being epistemologically grounded in experience.

19

Gautamı¯ya Vr: tti on 1.1.40; see Nyaya-Tarkatirtha and Tarkatirtha (1936, p. 326).

123

N. Guha

others. It seems every rational being is endowed a priori with a list of undesired things. I would argue that a priority is an important feature of tarka.

Avijn˜a¯tatattve ’rthe, Vibha¯ga and Ka¯ran: opapatti Kang wanted to understand tarka in terms of these above mentioned features. Let us review Kang’s understanding of these features. Kang (2010, p. 6) writes: ‘‘Since the object of tarka is not fully known in detail, the given representation can be neither completely confirmed nor disproved; this is a necessary precondition for the application of tarka.’’ This indeed is an important feature of tarka. When an epistemic instrument presents an object with its special properties and the presentation bears certainty (nis´caya), then no tarka is needed. In this case, the object is vijña¯tatattva. But a tarka is needed only when one has a doubt as to whether the object under investigation has a certain property or not. In this case, the object is under-informed or avijña¯tatattva. A tarka’s job is to validate one and only one option. The second important feature, according to Kang, is vibha¯ga which for him is ‘‘the differentiating or analyzing procedure’’.20 His argument is based on the following passages from Va¯tsya¯yana’s Bha¯s: ya: etasmim : s tarkavis: aye karmanimittam : janma iti prama¯n: a¯ni pravarttama¯na¯ni tarken: a¯nugr: hyante, tattvajña¯navis: ayasya vibha¯ga¯t tattvajña¯na¯ya kalpate tarka iti;21 and atha jijña¯sitasya vastuno vya¯hatau dharmau 22 vibha¯gena vimr: s´ati—kim : svid ittham a¯hosvin nettham iti. It seems Kang thinks that when epistemic instruments cannot generate certainty, a tarka begins by distinguishing the possible options involved in a case.23 According to Kang, in the second passage quoted above the distinction is made in the following way: Is it so or is it not so (kim : svid ittham a¯hosvin nettham iti)? This in other words means that ‘Which of the two options is true? Is X Y, or not?’ This physical analysis is vibha¯ga for Kang. But the traditional understanding of the term ‘vibha¯ga’ is quite different. Uddyotakara very clearly says: ‘‘What is vibha¯ga? It is the judgment regarding the logical fitness having the following form: This option is logically fit (yukta) whereas that one is not (ayukta).’’24 In fact, it is very likely that ‘vibha¯ga’ in the second passage (i.e.,. . .jijña¯sitasya vastuno vya¯hatau dharmau vibha¯gena vimr: s´ati) is nothing but showing the options separately. Here is Kang’s own translation of the passage:

20

Ibid, p. 5.

21

Rough Translation: In the cases such as ‘the regulator of birth’, some epistemic instruments get strengthened by tarkas. A tarka is used in order to know the true nature of an object, through categorizing the options that constitute the doubt. 22

The second passage has been translated in the section called ‘Tarka and Validation’.

23

Kang (2010, p. 8): Tarka is a means of sizing up a certain state of affairs: all the given factual or logical possibilities are differentiated, and the logical and theoretical implications of them are drawn.

24

kah: punar vibha¯gah: ? yukta¯yuktavica¯rah: , idam : yuktam idam ayuktam iti. Nya¯yava¯rtika on 1.1.1; see Nyaya-Tarkatirtha and Tarkatirtha (1936, p. 54).

123

Tarka as Cognitive Validator

Then the consideration in the form ‘‘Is this so or rather is it not so?’’ occurs, with a distinction being made between two [possible] contrary qualities of the object one desires to know.25 But we may notice that this distinction is not something tarka has made. It is made by a doubt (sam : s´aya). For here in this case the cognizing subject wants to know the object under investigation properly, and the following question arises in their mind: Is this so or not? The vibha¯ga made by tarka is rather a categorization. A tarka categorizes the options into two heads; logically fitting ones and unfitting ones. I do not find any reason to prefer Kang’s understanding to Uddyotakara’s. Let us now consider the term ‘ka¯ran: opapatti’. Va¯tsya¯yana says: ‘‘kim idam : janma kr: takena hetuna¯ nivartyate, a¯hosvid akr: takena, atha a¯kasmikam iti? evam avijña¯te ’rthe ka¯rn: opapattya¯ u¯hah: pravartate’’26 which roughly means the following: ‘Is birth brought about by a caused cause, or an uncaused cause, or it is accidental? As there is an uncertainty regarding the under-informed object, a cognitive act (u¯ha) starts operating through ka¯ran: opapatti’. Kang thinks that the term ‘ka¯ran: opapatti’ means postulating reasons for each of the three possibilities. And the job of furnishing reasons is assigned to a tarka.27 Vatsya¯yana himself says: sambhavati asmin ka¯ran: am : prama¯n: am : hetur iti [Translation: ‘It is logically fit to say that this option must be supported by an epistemic instrument’—this is ka¯ran: opapatti].28 We may see what Uddyotakara has to say about this term. He writes: ‘upapattih: sambhavah: ’. I think that sambhava is nothing but logical fitness or tenability. It is the antonym of asambhava or un-tenability or logical unfitness. I do not think that sambhava means ‘possibility’ in a statistical sense. It is not even ‘inference to the best explanation’. I would support my claim by examining a few schematic forms of tarka. Here is a reductio schema (Indirect Method): 1. 2. 3.

Z is a case of X Any case of ~ Y is a case of ~ X Z is a case of ~ Y

(Premise) (Premise) (Assumption)

4. 5. 6.

Z is a case of ~ X Z is a case of both X and ~ X Therefore, Z is not a case of ~ Y

(From 2 and 3) (From 1 and 4)

25

See Kang (2010, p. 7).

26

Nya¯yabha¯s: ya on 1.1.1; see Nyaya-Tarkatirtha and Tarkatirtha (1936, pp. 53–54). Kang (2010, pp. 4–5):

27

The passage in the NBh ‘‘evam avijña¯tatattve ’rthe ka¯ran: opapattya¯ u¯hah: pravartate’’ following the one example of tarka (tasyoda¯haran: am : ‘‘kim idam : janma kr: takena hetuna¯ nirvartyate, a¯ho svid akr: takena, atha¯kasmikam’’ ity) calls for special attention. In this passage ka¯ran: opapattya¯ is in the instrumental case, indicating that u¯hah: proceeds by furnishing persuasive reasons (ka¯ran: opapattya¯). This sentence suggests that the following consideration of three possibilities, each backed by reasons, means that tarka, too, proceeds by furnishing persuasive reasons (ka¯ran: opapattya¯). According to this interpretation, the ‘‘ka¯ran: opapatti’’ in NS 1.1.40 refers to the reasons furnished for each possibility. Provided that this interpretation is justified, the passage ‘‘evam avijña¯tatattve ’rthe ka¯ran: opapattya¯ u¯hah: pravartate’’ implies that u¯ha, taken as the content of tarka, is generated by consideration of the reasons postulated for each possibility. 28

See the beginning of the section, ‘Tarka and Validation’.

123

N. Guha

This schema will generate only true conclusions provided that the premises are true. Epistemological contents may replace the variables in the premises. If the epistemic instruments that bring forth the epistemic content are defective, then tarka too will fail to generate true conclusions. A reductio tarka of the aforementioned type begins by assuming the denial of each option. The denial of the tenable one does lead the argument to a contradiction. No other denial does so. This is the upapatti or sambhava. Why is this ka¯ran: opapatti? Every commentator says that here ‘ka¯ran: a’ means an epistemic instrument (prama¯n: a). We may consider the following case. ‘Is birth brought about by a caused cause, or an uncaused cause, or is it accidental?’ In this case, a doubt consists of three options. The Naiya¯yika would say that here, only the first option is logically fit. Any other option will lead the arguer to an undesired factor.29 A tarka is the cognitive act of appreciating the fitness or un-fitness of a given option. Validation is nothing but assigning epistemic validity to an option O. Now ‘O is valid (pra¯ma¯n: ika)’ means that if one could directly check this case, then one would have cognized that ‘O is the case’ through an epistemic instrument (prama¯n: a). O could have been brought forth by an epistemic instrument. One who knows that fire causes smoke argues: ‘any case of smoke is a case of fire too; for an effect could not exist without its cause’. The supporting reductio tarka here suggests that if one could check all cases of smoke, then one would have perceptually known that those are cases of fire too. The reductio tarka tells one that the option that any case of smoke is a case of fire (i.e., fire pervades smoke) must be supported by an epistemic instrument. According to Uddyotakara, a reflective argument (u¯ha) can re-apprehend that which has already been known through an epistemic instrument. Suppose one knows through some epistemic instrument that P is true. Still one may conclude that P by using an u¯ha and think ‘that which I came to know through an epistemic instrument has been ascertained now by using a counterfactual reasoning’. But in such a case, this u¯ha is not a tarka.30 That means an u¯ha is an argument that leads one to a valid conclusion, provided the other cognitions involved in it are valid.

Definition of Tarka Tarka has already been implicitly defined in this paper. In this section, the intended definition will be uttered explicitly. I think it is true that tarkas are based on a priori principles. But this statement just tries to understand tarka from the western

29 Va¯tsya¯yana has shown how the second and the third option will lead one to an undesired factor. He says:

yadi kr: takena hetuna¯ nivartyate, hetu¯ccheda¯d upapanno ’yam : janmocchedah: . atha¯kr: takena hetuna¯, tato hetu¯cchedasya as´akyatva¯d anupapanno janmocchedah: . atha¯kasmikam : , ato’kasma¯n nivartyama¯nam : na punar nivartsyatı¯ti nivr: ttika¯ran: am : nopapadyate, tena janma¯nuccheda iti. See Nyaya-Tarkatirtha and Tarkatirtha (1936, p. 54). I am not translating this passage here. 30 ubhayatha¯ khalv ayam u¯ho bhavati vijña¯tatattve ’vijña¯tatattve ca. . .yo ’sa¯v arthah: pu¯rvam anena vijña¯tas tam evohata iti pu¯rva¯nubhu¯taparicchedaka u¯ha ity ucyate, na tv ayam : tarkah: . Nya¯yava¯rtika on 1.1.1; Ibid, pp. 324–325.

123

Tarka as Cognitive Validator

perspective. What is tarka in the framework of Old Nya¯ya? Here is my formulation of the definition of tarka: yena anis´ca¯yakena u¯hena avijña¯tatattve ’rthe tattvajña¯na¯rtham : ka¯ran: opapattya¯ ekasya sambha¯vitadharmasya abhyanujña¯, aparasya anis: :tavya¯pyasya vyuda¯so va¯ sa tarkah: . Translation: A tarka is a reflection (i) which does not generate certainty, (ii) which is applied in order to know whether an under-informed object has a specific property or not, and (iii) which validates just one option by demonstrating its logical fitness and/or demonstrates that the rivaling option is unfit since it would lead the cognizing subject to something undesired. Why should the Naiya¯yika say that a tarka cannot generate certainty?31 The reason is this: A tarka lacks any epistemological content. It just demonstrates the logical fitness of an epistemic instrument. Va¯caspati says: ‘The demonstration of the logical fitness cannot generate certainty alone; for it cannot establish itself in the absence of a shelter (a¯s´raya)’.32 A tarka which has nothing but an abstract form has to be sheltered in an epistemic instrument and enhance its host. A tarka just destroys the blocker of the certainty of an epistemic instrument. Tarka rejects an undesired content that belongs to a cognition or a speech act (such as a statement or a question). Profligacy and contradiction are undesired. Tarka is a cognitive act that points to profligacy in a cognition or contradiction in a statement. I shall show later that vaiya¯tya tarka demonstrates the undesired-ness of the content of a question. One may doubt whether asking such a question is valid or not. When the undesired-ness of a content is pointed out by a tarka, the content is rejected. I think that the Naiya¯tika defines tarka in terms of its incapability to generate certainty (nis´caya¯yogyata¯) for a very significant reason. Without this qualification, the definition would have been over-generalized. Suppose the cognizing subject sees something that is five and a half feet tall and doubts whether it is a man or a post. This is certainly a case of avijña¯ta-tattva. This doubt can be dispelled by a perceptual cognition, viz., ‘This is a man’. But the second cognition (or whichever epistemic instrument has generated it) is not a tarka. A tarka has an abstract general form and nothing else whereas a cognition has an epistemic content.

_ ´a’s Application of Tarka Ganges _ ´a, the primary meaning of tarka is the reductio argument. But fortuFor Ganges nately, he considers the weak tarkas such as the law of parsimony, default value arguments etc. as tarkas. In the following section, I shall try to critically examine his _ ´a uses tarkas of different types for supporting his own applications of tarka. Ganges position and refuting others’ positions. He has made many significant comments on _ ´a’s applications and comments to exemplify and tarkas. We shall use Ganges understand tarkas. 31 32

na tv avadha¯rayati. Nya¯yava¯rtika on 1.1.1; Ibid, p. 54. utpatteh: svatantra¯ya¯ a¯s´raya¯siddhataya¯ svato nis´caya¯yoga¯t. Tatparya-t:¯ıka¯ on 1.1.40; Ibid, p. 321.

123

N. Guha

Reductio, Contradiction and Infinite Regress I shall first roughly define these tarkas. Reductio (tad-anya-ba¯dhita¯rtha-prasan_ga): Davis (2001, p. 157) presents the general schema of this tarka as follows: 1 Pa 2 (x) (Px  Qx) 3 Qa 4 ~Qa Therefore, ~Pa

assumed premise premise

We can now instantiate this schema. Suppose the cognizing subject S has a doubt as to whether fire pervades smoke (i.e., any locus of smoke is a locus of fire) or not. Suppose S knows that smoke is a cause of (i.e., a necessary condition for) fire. It is evident that if X can exist without Y then X is not caused by Y. In order to dispel the aforementioned doubt S argues in the following way: (1) Smoke can exist without fire. (Assumption) (2) Anything that can exist without fire cannot be caused by fire.33 (Premise since this is evident) (3) Smoke is not caused by fire. (From 1 and 2) (4) Smoke is caused by fire. (Premise since S knows this) Therefore, smoke cannot exist without fire, i.e., fire pervades smoke. (3) and (4) are contradictory. So the assumption (viz., smoke can exist without fire) must be false. In that case the denial of the assumption, i.e., smoke cannot exist without fire, i.e., any locus of smoke is a locus of fire, i.e., fire pervades smoke, is true. Contradiction (vya¯gha¯ta): ‘That which is existent is not there’,34 ‘Here is a round square’—these are contradictions. If it is known that P, and somebody argues that it is not the case that P, then their argument too involves contradiction or _ ´a talks about a vya¯gha¯ta. Not only logical contradictions are contradictions. Ganges behavioural contradiction. I have already given an example of Infinite Regress (anavastha¯) towards the end of the section called ‘Strong and Weak tarkas’.

33 I have deliberately avoided the conditional, i.e., (x) (Px  Qx). I believe that pervasion is the inclusion of a set into another. Y pervades X if and only if any locus having X has Y. Thus pervasion is understood as ‘coexistence’ (sa¯ma¯na¯dhikaran: ya) which is a special case of ‘existence’. In that sense, a statement of pervasion is an ‘existential’ statement whereas implication does not guarantee ‘existence’. That is why the statement, ‘for all x, if x is a unicorn, then x is white’ is (vacuously) true; but the statement, ‘any case of unicorn-ness is a case of whiteness’ or ‘whiteness pervades unicorn-ness’ is false. According to the Naiya¯yika, it will always generate a fallacious inference, since unicorn-ness or the property of being a unicorn does not exist anywhere (aprasiddha). 34 san na¯sti. Khan: d: anakhan: d: akha¯dya by S´rı¯ Hars: a, Chapter IV; see Vaija¯purakara and S´a¯strı¯ (1962, p. 560).

123

Tarka as Cognitive Validator

Text: Tattvacinta¯man: i (Tarka: The Chapter on Tarka) T-1. tatha¯ hi dhu¯mo yadi vahnyasamavahita¯janyatve sati vahnisamavahita¯janyah: sya¯n notpannah: sya¯d ity atra, kim : dhu¯mo ’vahner eva bhavis: yati, kvacid vahnim vina pi bhavis yati, ahetuka eva vyotpatsyata iti s´an_ka¯ sya¯t. sarvatra ¯ : : svakriya¯vya¯gha¯tah: sya¯t. yadi hi gr: hı¯ta¯nvaya-vyatirekam : hetum : vina¯pi ka¯ryotpattim : vahneh tr ptyartham s´an_keta tada¯ svayam eva dhu¯ma¯rtham : : : : bhojanasya parapratipattyartham : s´abdasya copa¯da¯nam : niyamatah: katham : kurya¯t, tena vina¯pi tatsambhava¯t.35 [Here is a tarka:] Had smoke been different from both that which is caused by fire and that which is not caused by fire, it would not be produced (utpanna) at all. These are the doubts [that are to be dispelled by the aforementioned tarka]: (1) Whether smoke is caused only by elements that are different from fire or not; (2) Whether there can be any smoke without fire or not; (3) Whether smoke is uncaused or not. [If these doubts remained] then they would have opposed one’s own behaviour. One hypothesizes that X causes Y by observing that X and Y are related through causal agreement and difference (anvaya-vyatireka). If an effect E could take place without the factor F that is related to E through causal agreement and difference, then the following activities could not be accounted for: one who needs smoke lights a fire; in order to satisfy one’s hunger one eats; in order to convey a message one speaks. If Y could take place without being causally related to X, then one should not go for X when one looks for Y. Comment: Suppose S has observed that X regularly precedes Y, and Y cannot occur in the absence of X. This means that X and Y are related through causal agreement and difference (anvaya-vyatireka). After having observed this bi-polar relation between X and Y, S is supposed to hypothesize that X causes Y. But S still has the following doubts: (1) Whether Y is caused by non-X only; (2) Whether there can be Y without X; (3) Whether Y is uncaused or not. These doubts are to be dispelled by the following tarka: Had Y been different from both that has been caused by X and that has not been caused by X, then Y would not have been produced at all. But suppose S knows that Y is produced. S cannot dismiss a piece of knowledge. Thus this tarka will insist S to accept that Y is caused by something since Y is produced. This is how doubt (3) (i.e., ‘whether Y is uncaused’) is dispelled. Now that ‘something’ must be X since it has been observed that X and Y are related through causal agreement and difference. Thus doubt (1) (i.e., ‘Whether Y is caused by nonX only’) is dispelled since S comes to know that Y cannot be caused by non-X only. Thus S concludes that X causes Y. Finally what remains is doubt (2) (i.e., ‘Whether there can be Y without X’). While discussing reductio tarka, I have already mentioned how this doubt could be dispelled by the following tarka: If Y (smoke) could _ ´a has exist without X (fire), let Y (smoke) not be caused by X (fire). Thus Ganges demonstrated how a single tarka could possibly dispel a sequence of doubts.

35

See Tarkavagish (1990, pp. 219–229, Volume II, Part 1).

123

N. Guha

T-2. tasma¯t tattadupa¯da¯nam eva ta¯dr: s´as´an_ka¯pratibandhakam. s´an_ka¯ya¯m : na 36 niyatopa¯da¯nam : , niyatopa¯da¯ne ca na s´an_ka¯. The very fact that one goes for X when one needs Y opposes the doubt as to whether X causes Y or not. When one has such a doubt one is not convinced of the causal relation between two factors, and when one has such a conviction one cannot have the aforementioned doubt. Comment: S may have a doubt as to whether every case of Y is a case of X. In that case, S should look at the behaviours of people. When one needs smoke one lights a fire. When one is hungry, one eats. When one wants to say something, one talks. These activities or general behaviours are consistently successful because every case of Y (smoke or satisfying one’s hunger or communicating with others) is indeed a case of X (fire or eating or using linguistic units). In general, a doubt, ‘whether a or not’ is contradictory to either the certain cognition, ‘a’ or to the certain cognition, ‘it is not the case that a’. Thus a doubt having the form, ‘whether each case of Y is a case of X too’ is contradictory to the doubt-free cognition, ‘any case of Y is a case of X’. As long as one has the former, one cannot have the latter. Conversely had one been fully convinced that any case of Y is a case of X, one could not have had the doubt as to whether each case of Y is a case of X. The general successful behaviours show that nobody doubts the pervasion between smoke and fire etc. Therefore one should convince oneself that fire pervades smoke. T-3. etena vya¯gha¯to virodhah: , sa ca saha¯navastha¯naniyama iti tatra¯py anavastheti 37 nirastam : . svakriya¯ya¯ eva s´an_ka¯pratibandhakatva¯t. One could say that: ‘Contradiction (vya¯gha¯ta) is an opposition (virodha). It is nothing but a rule having the following form, ‘that which is a case of being X cannot be a case of being Y and vice versa’. [This is pervasion of some kind. In order to know this rule or pervasion one has to invoke another tarka; for in order to ascertain a pervasion some tarka is needed.] Thus there would be infinite regress (anavastha¯).’ But this view is not tenable, since one’s own [successful] activities [based on one’s convictions] would destroy such doubts. Comment: Here is a rough translation of a passage from Khan: d: anakhan: d: akha¯dya by S´rı¯ Hars: a. Suppose S has a doubt as to whether X pervades Y. In order to dispel this, S applies the following tarka: If Y could exist without X, then let X not cause Y. The heart of this tarka is the pervasion, ‘anything that is caused by X does not exist without X’. Suppose S starts doubting the pervasion between being-caused-by-X and not-existing-without-X. S needs to invoke another tarka here and ad infinitum. Thus it leads S to infinite regress (anavastha¯). One may say that vya¯gha¯ta or behavioural contradiction ends the series of doubts. But this position too would be untenable; for the heart of the behavioural contradiction is this pervasion: One who has a doubt as to whether Y (fire) pervades X (smoke) cannot go for Y when one needs X. [We may remember here that one who needs smoke lights a fire.] What if S doubts this very pervasion? Then in order to dispel the doubt S needs another tarka. Thus infinite 36

Ibid, pp. 229–231.

37

Ibid, p. 232.

123

Tarka as Cognitive Validator

_ ´a is answering the aforementioned objection regress once again!38 Now Ganges _ ´a says: raised by S´rı¯ Hars: a. In the chapter on the apprehension of pervasion, Ganges ‘One applies tarkas as long as one has a doubt. When a behavioural contradiction puts an end to the series of doubts, no tarka is needed any more’.39 This answers the first part of S´rı¯ Hars: a’s objection. Maybe in a hypothetical situation, S will doubt everything. But supposing that S is a real subject with real knowledge of real objects, we may say that S’s doubt about the pervasion between X and Y will cease to exist the moment S will reach a behavioural contradiction. Now S´rı¯ Hars: a would ask his second question: Will the chain of doubt really cease to exist at the emergence of the understanding of a behavioural contradiction? The behavioural con_ ´a would once tradiction itself is based on a pervasion which can be doubted. Ganges again say that being a real subject, S would stop doubting when S reaches a behavioural contradiction. Such is the epistemic nature of S. Let there be a pervasion at the heart of the contradiction. The point is that the very knowledge of the _ ´a is merely contradiction or vya¯gha¯ta tarka will dispel all the doubts. Here Ganges describing what happens to a cognizing subject in such cases. Economy and Profligacy Arguments based on economy and profligacy are somewhat similar to Ockham’s Razor or the principle of lex parsimoniae.40 In the section on postulation _ ´a says: (Artha¯pattiva¯da), Ganges The tarkas based on economy and profligacy (la¯ghava-gaurava-tarka) do not come to a stop in contradiction (viparyaya-aparyavasa¯yin). Nevertheless they function as auxiliary causes for valid cognitions. Therefore, even in perceptual or testimonial cognitions, they serve as auxiliary causes.41 That means an economy argument (la¯ghava-tarka) is a tarka even though it does not end in contradiction. What does it do then? This is answered in the following section of Tattvacinta¯man: i. Text: Tattvacinta¯man: i (I¯s´varava¯da: The Chapter on God) T-4. na ta¯vad anumitima¯tre la¯ghavam : sahaka¯ri. . .yatra prama¯n: e laghuguruvis: ayata¯ sambhavati, tatra la¯ghavasahaka¯rita¯. ka¯ran: ata¯ka¯ryata¯vya¯pyatva¯digra¯hake pratyaks: e pravr: ttinimittagra¯hake upama¯ne s´abdas´aktigra¯haka¯numa¯ne tatha¯vidhaprama¯n: ama¯tre 38

atha tarkasya vya¯ptir mu¯labhu¯ta¯ ’bhyupagamyate. tatra¯pi vyabhica¯ras´an_ka¯ya¯n_ punar anavasthaiva. tatra¯pi vya¯gha¯ta¯pa¯dane punar ittham anavasthaiva. Khan: d: anakhan: d: akha¯dya by S´rı¯ Hars: a, Chapter I; see Vaija¯purakara and S´a¯strı¯ (1962, p. 271). 39 yavada¯s´an_kam : tarka¯nusaran: a¯t. yatra ca vya¯gha¯tena s´an_kaiva na¯vatarati tatra tarkam : vinaiva vya¯ptigrahah: . Vya¯ptigrahopa¯yasiddha¯nta (The chapter of the apprehension of pervasion); Tarkavagish (1990, pp. 210–212, Volume II, Part 1). 40 The popular Latin version of this is: Numquam ponenda est pluralitas sine necessitate (entities should not be multiplied unnecessarily). 41

na ca va¯cyam : tarka¯n: a¯m : viparyaya¯paryavasa¯yitve a¯bha¯satvam. tatparyavasa¯ne ca tad eva¯numa¯nam etattarkasaha¯yam : bahih: sattvam anuma¯payis: yatı¯ti. yato la¯ghavagauravatarka¯n: a¯m : viparyaya¯paryavasa¯yina¯m eva prama¯n: asahaka¯ritvam, ata eva pratyaks: as´abda¯dau api sahaka¯rı¯ sah: . See Tarkavagish (1990, 655–656, Vol. II, Part 1).

123

N. Guha

sakalata¯ntrikaih: sahaka¯ritvakalpana¯t. evam : la¯ghavam evam : gauravam iti jña¯na¯nantaram : ba¯dhakam : vina¯ laghu¯na¯m eva ka¯ran: atvaka¯ryatvavya¯pyatvapravr: ttinimittat42 vas´abdas´akyatva¯na¯m : jña¯nadars´ana¯t. It is not true that every inference is supported by an economy tarka. . .Where there is a possibility of two or more cognitions and a piece of cognition is more economical than others, only there such a tarka is an auxiliary cause for the valid cognition. Every philosopher accepts that the perceptual cognitions of cause-ness, effect-ness, pervaded-ness etc., the analogical cognition that enables one to know the meaning of a word, the inferential cognition that enables one to know the semantic relation between a word and its referent, and many such valid cognitions have economy tarkas as auxiliary causes. For everybody intuits, ‘this is economical and that is not economical’, and then accepts the most economical option [as the valid one]; provided the most economical option is not contradicted [by any other valid cognition]. Comment: A number of possible cognitions may compete with each other for epistemic validity. But only the one supported by an economy tarka finally wins. Here the law of parsimony or lex parsimoniae operates as the selection criterion. The idea is this: In a formative stage, when several cognitions rival each other, the most economical one is supported by the principle of economy and wins the epistemic game. The winning moment marks the transition to a valid cognition. Before winning, the cognition is not valid, but just a cognition, often unrecognized. Suppose in the model M there are smoke-individuals of three kinds; each one foot high, two feet high and three feet high. In M, fire pervades smoke, i.e., every case of smoke is a case of fire. Raw observation has provided the cognising subject S with smoke-individuals of all three heights. S could have cognized that ‘fire pervades all one foot high smokes, and it pervades all two feet high smokes, and it pervades all three feet high smokes’. But S happens to cognize that fire pervades smoke in general. The job of the economy-based argument is to ignore the factor of height and consider smoke-in-general alone. Pervaded-ness (vya¯pyatva) is perceived in smoke in general, since conceptually ‘smoke’ is more economical than the conjunction of ‘one foot high’, ‘two feet high’, and ‘three feet high smoke_ ´a, everybody intuitively understands which individuals’.43 And according to Ganges option is economical and which is not (evam : la¯ghavam evam : gauravam iti ´ jña¯na¯nantaram . . . jña nadars ana t. See T-4). After economy is appreciated, there ¯ ¯ : arises the valid cognition. In the genesis of any valid cognition some of the factors are ignored and some are emphasized by lex parsimoniae. In the case of the pervaded-ness of smoke, the height of smoke is ignored and smoke-ness is emphasized. _ ´a can be elaborated here. He says that la¯ghava Another example given by Ganges tarka plays a role in the genesis of ‘the inferential cognition that enables one to know the semantic relation between a word and its referent’. Such a semantic relation is called ‘s´akti’. The process of iterated learning in the Indian framework is 42 _ ´a’s ¯Is´varava¯da in Tarkavagish (1990, pp. 76–78, Vol. II, Part 2). Vattanky’s translation of Ganges Vattanky (1984, pp. 235–236) has helped me translate this section. 43

If Y pervades X, then Y is the perdader (vya¯paka) and X the pervaded (vya¯pya). That means that here X and Y have pervaded-ness and pervader-ness respectively.

123

Tarka as Cognitive Validator

the following. The speaker asks the hearer to bring a cow. The hearer brings a cow. Then the speaker says, ‘tie the cow’. The hearer does so. The learner observes the whole thing and compares the pair of utterances to the pair of activities. The expression ‘the-cow’ is common in both the utterances and there is a cow involved in both the activities performed by the hearer. Therefore ‘the-cow’ must refer to a cow. The expressions ‘bring’ and ‘tie’ must mean the activities of bringing and tying respectively. There is an inference at the centre of the whole process of language learning. How did the learner know that different parts of the expression, ‘bring me the cow’ refer to bringing, cow etc.? The learner must have inferred that nothing but the utterance of the speaker has triggered the activity of the hearer. Here is the sequence of inference: ‘The activity [of the hearer] (kriya¯) is caused by an effort (kr: ti). The effort is caused by a cognition. That cognition is caused by a sentence (va¯kya)’.44 Thus the learner infers that the activity of the hearer is causally linked with a linguistic utterance. This inference is based on economy since no other causal link would have been any more economical than this one. Default Value and Specific Value Here is a default value argument (utsarga tarka) cited by S´rı¯ Hars: a: ‘Suppose A is somebody whose senses are working fine and who is awake. A has the cognition C. There is no evidence that C is true. But still there is no evidence against C. In such a case, if another person B says that C is not true, then B’s argument will be countered by the default value argument; for somebody with perfect senses normally has true cognitions. If somebody says that A’s cognition was true then their argument will not be countered by the default value argument.’45 Text: Tattvacinta¯man: i (S´aktiva¯da: The chapter on semantic relation) T-5. ata eva pra¯ma¯n: yasya autsargikatve ’pi ba¯dhaka¯bha¯vasahakr: tanis´caya¯d eva 46 nis´cayah: , anyatha¯ prama¯prama¯ va¯ . . . iti sam : s´ayah: kva¯pi na sya¯t. One may assume that S’s cognition C is valid (pra¯ma¯n: ika) by default. But the default condition for the certainty of the validity of C should be the following: There must be a doubt-free cognition that ‘there is no factor that could invalidate C’. Otherwise nobody could have the doubt as to whether their cognition is valid or not. Comment: Suppose somebody with perfect senses, sees something in Sundarban (the famous forest in the vast delta on the Bay of Bengal) and cognizes that ‘there is a snake in the forest’ (¼C1). C1 should not be doubted or considered invalid unless there is strong evidence against it. On the contrary, suppose somebody from Sundarban with defective eyes sees something in the middle of Berkeley Square, 44

iyam : kriya¯ kr: tijanya¯, sa¯ jña¯najanya¯, tad va¯kyajanyam ityanumitiparamparaya¯. . . Ka¯rya¯nvitas´aktiva¯dasiddha¯nta (The chapter on Semantic Relation); See Tarkavagish (1990, pp. 499–500, Volume IV, Part 2). 45 svasthasya ja¯grato jña¯nam : pra¯ma¯n: ya¯pra¯ma¯n: yanirdha¯rakaprama¯n: a¯nupanipa¯ta¯vis´es: e ’pi vina¯ ba¯dham apra¯ma¯n: yam abhyupagacchantam : prati sya¯t, na tu pra¯ma¯n: yam. Khan: d: anakhan: d: akha¯dya by S´rı¯ Hars: a, Chapter IV; see Vaija¯purakara and S´a¯strı¯ (1962, p. 568). 46

See Tarkavagish (1990, pp. 497–498, Volume IV, Part 2).

123

N. Guha

London on a cloudy day and cognizes that ‘that is a King Cobra’ (= C2). Nobody else saw any snake there. Now there is strong evidence against C2 since there is perhaps no King Cobra in the UK, and above all S finds that there was a rope in the middle of Berkeley Square. The set of factors that suggests that what that person from Sundarban saw was perhaps not a snake is the invalidator (ba¯dhaka) for C2. If there is an invalidator for a cognition then the cognition can always be doubted. But if there is no such factor, there is no reason to doubt the validity of a cognition. Therefore, every certain cognition should be considered valid by default. It should be doubted only when there is an invalidator.

Concluding Remarks We may remember that there are tarkas of two types; strong tarka or the tarka based on ‘undesired counterfactual imposition’ (prasan_ga¯tmaka tarka) and weak tarka or the tarka that is not based on any ‘counterfactual imposition’ (aprasan_ga¯tmaka tarka). Perhaps we may add ‘contradiction’ (vya¯gha¯ta) to the list of strong tarkas since it is the very essence of prasan_ga. I have already discussed weak tarkas of four types, namely economy (la¯ghava), profligacy (gaurava), default value (utsarga) and specific value (apava¯da). We may also notice here that economy-andprofligacy and default-value-and-specific-value are pair tarkas (yugma-tarka). Something cannot be economical unless something else is profligate. The question of a default value does not arise had there been no exception. I may introduce another weak tarka. Kang makes an interesting comment: Such questions, depending on the interpretation and analysis of them, may lead to different kinds of mainly non-formal fallacy: the fallacy of many questions (‘Are you still beating your wife?’), the false dilemma (‘Are you a good man or a bad man?’) or a category fallacy (‘How many iron filings does a unicorn attract?’). When such a question has been put forward for a tarka, an argumentative or epistemological quandary is the guaranteed outcome, at least from the Buddhist standpoint. Any reasonable analysis of the logical or factual implications is beside the point, and a pseudo-analysis results. In such a way, tarka can become an easily misused quasi-formal method.47 I would like to respond to this point raised by Kang. Such questions cannot even be put forward for a tarka, since they will be rejected by a (weak) tarka, namely vaiya¯tya. This is S´rı¯ Hars: a’s definition of vaiya¯tya: ‘pra¯ma¯n: ika¯vyavaha¯ryatvam asama¯dheyaja¯tı¯yam anaucityam : vaiya¯tyana¯makam’ which may be roughly translated as this: ‘The tarka that dismisses the content of those questions etc. whose answers cannot be epistemologically valid or whose answer is impossible is a vaiya¯tya’.48 S´rı¯ Hars: a’s examples are much more interesting than his definition. The first example is: ‘avastuni vidhinis: edhayoh: kim icchasi’.49 Somebody who thinks 47 48 49

Kang (2010, p. 21). Vaija¯purakara and S´a¯strı¯ (1962, p. 470). Ibid, p. 471.

123

Tarka as Cognitive Validator

that everything is ontologically void, empty (s´u¯nya) asks a realist, ‘Is a non-entity positive or negative?’ A positive entity is that which has a positive existence, e.g., a pot. A negative entity is the absence of pot. The absence ‘exists’ since there is no pot. The realist does not think that everything is ontologically void. In this context the most important point is this: whichever way the realist will answer the question would assume that everything is void. Void-ness is undesired for the realist. This is surely a fallacy of many questions. The second example given by S´rı¯ Hars: a is: 50 ‘apratı¯te davadattau sa kim : gaurah: kr: s: n: o veti’. Devadatta is not known to anybody. Still one asks somebody, ‘Is Devadatta fair or dark in complexion?’ This is a category fallacy. Vaiya¯tya tarka dismisses both the questions by showing that they involve something fallacious. Thus we may say that if such a question has been put forward for a tarka, there would be no argumentative or epistemological quandary; in fact it is to be rejected by a tarka. It seems that Kang agrees with somebody who says that tarka is an ‘‘unreliable speculation’’. For him, it is unreliable since ‘‘it cannot offer us a definitive solution to the problem at hand’’ and can be misleading too.51 But if a tarka is unreliable, so is every logical deduction rule such as Modus Ponens. A logical deduction will yield a true conclusion, provided that the premises are true. If a tarka is misleading, so is an epistemic instrument such as an eye; for the latter may mistake something for something else. An epistemologist who has no complaints against logical deduction rules or epistemic instruments should not have any problems with tarka. Strong tarkas are like deduction rules that guarantees logical consistency. I may add another point here. Most of the applications of weak tarkas (such as economy or default value) are involuntary. One sees that a white cow has a dewlap, a black cow has a dewlap . . ., and cognizes that ‘any cow has a dewlap’. This cognition automatically overlooks all the irrelevant differences. Likewise, if there is no counterevidence, one accepts by default what a normal person experiences. Thus default value plays a great role in epistemology. One cannot help these epistemological behaviours. When a tarka is used voluntarily, it in the most cases acts as a metatheory, a criterion for accepting or rejecting a theory. In such cases, these tarkas are extensions of automatic human epistemological behaviours. For example, one _ ´a has already deliberately uses economy arguments to evaluate a theory. And Ganges shown how an economy tarka was used in the involuntary processes of cognitionformation. One of the jobs of an epistemologist is to describe the epistemological behaviour of human beings correctly and to enlist the conditions under which a cognizing subject thinks that their cognition is valid. Tarka is one such condition 50

Ibid, p. 47.

51

Kang (2010, p. 21). Once again Kang (2010, p. 21): However, this method can be utterly misleading. The universal applicability of tarka is its strong point, but at the same time it is a serious drawback. For any given object we can suggest alternative qualities seemingly applicable to that object, and ask which alternative is the suitable one. . . . The fact that it does not stand or fall on the basis of factual information was, I surmise, the reason for the widespread negative connotation of the term ‘tarka’. The widespread reluctance against tarka in ancient Indian intellectual traditions including Buddhist disapproval of tarka as unreliable speculation could be understood on the bases of the analysis discussed in the present paper.

123

N. Guha

which accounts for many epistemological phenomena. That is why Nya¯yasu¯tra 1.1.40 says that it is a reflection that functions in order to know the truth (tattva¯vadha¯ran: a¯ya) of an object. Here, tattva is ‘‘‘the condition that allows for valid representation [of the object] in thought and speech,’ in the epistemological context of Nya¯ya’’.52 Under some conditions, the visual cognition ‘this is water’ is considered valid. Likewise, when one applies a tarka in order to know whether the under-informed object O has a specific feature F, and the tarka tells one that O has F, one considers the cognition ‘O has F’ as a valid one. As far as the objects that are represented by an epistemic instrument are concerned, their representations must conform to their truth. But as far as the objects that are known through a tarka are concerned, their truth must conform to their representations. If somebody can do away with all epistemological tools, then they can sacrifice tarka too. But for somebody, who thinks that epistemic instruments represent objects as they are, sacrificing tarka would be like sacrificing the baby for the bath-water. For the Naiya¯yika can always attribute the defect of the conclusion of a tarka either to an epistemic instrument or to an absence of epistemic instruments. Acknowledgments I acknowledge my great indebtedness to Prof. Chakravarthi Ram-Prasad (Associate Dean for Research, Faculty of Arts and Social Sciences, Lancaster University), my PhD supervisor, for making valuable comments on earlier drafts of this paper. I also thank Dr. Anindita Chakrabarti (Assistant Professor, Department of Humanities and Social Sciences, Indian Institute of Technology, Kanpur), for her suggestions.

References Davis, L. (2001). Tarka in Nya¯ya theory of inference. In R. W. Perrett (Ed.), Indian philosophy: A Collection of readings (Vol. 2). New York: Garland Publishing, Inc. Dravida, L. S., & Dvivedin, V. P. (Eds.). (1911). Nya¯ya-va¯rtika–ta¯tparya-paris´uddhi by Udayana¯ca¯rya with a gloss called Nya¯ya-nibandha-praka¯s´a by Vardhoma¯nopa¯dhya¯ya. Calcutta: Asiatic Society of Bengal. Kang, S. Y. (2010). An inquiry into the definition of tarka in Nya¯ya tradition and its connotation of negative speculation. Journal of Indian Philosophy, 38(1), 1–23. Matilal, B. K. (1986). Perception: An essay on classical Indian theories of knowledge. Oxford: Clarendon Press. Matilal, B. K. (1998). In J. Ganeri & H. Tiwari (Eds.), The character of logic in India. Albany: State University of New York Press. Mohanty, J. N. (2000). Classical Indian philosophy. Lanham, MD: Rowman & Littlefield Publishers, Inc. Nyaya–Tarkatirtha, T., & Tarkatirtha, A. (Eds.). (1936). Nya¯yadars´anam. Calcutta: Metropolitan Printing & Publishing House. Sharma, U. S. (Ed.). (1978). S´rı¯manma¯dhava¯ca¯ryakr: tah: Sarvadars´anasam : grahah: . Varanasi: The Chowkhamba Vidyabhavan. Tarkavagish, K. (Ed.). (1990). Tattvacinta¯man: i of Gan_ges´a. Varanasi: Chaukhamba Sanskrit Pratishthan. Vaija¯purakara, G., & S´a¯strı¯, S. (Eds.). (1962). Khan: d: anakhan: d: akha¯dya of S´rı¯ Hars: a. Ka¯s´¯ı: Acyuta Granthama¯la¯. Vattanky, J. (1984). Gan_ges´a’s philosophy of god. Madras: The Adyar Library and Research Centre.

52

Ibid, pp. 5–6.

123