The buddhist theory of inference (anumana)
The epistemological thinkers in india have generally adopted a causal approach to knowledge. knowledge is taken to be an occurrence, an outcome of a particular causal complex (karana samagri) in which the causal condition acting as an instrumental cause (karana) is known as pramana. pramana is the mode of knowing. the buddhist thinkers do not entertain the distinction between pramana and its outcome (pramana phala= prama) mainly because this distinction is not needed in their epistemological set up. as opposed to the school of nyaya which maintains such a distinction because of its presupposition that pramana is the ground for the truth of a prama (manadhina meyasiddhih) which is its phala (outcome), the buddhists repudiate this distinction because for them a reference to the object of knowledge (prameya) is the ground for the truth of prama.1 moreover, they maintain that no rigid separation is possible between the act of cognizing and the cognition of the object.2
anumana, which in the buddhist tradition is one of the two pramanas, is at once a mode of knowing and a way of reasoning. thus it has an epistemic as well as a logical aspect.
the word anumana literally means ‘a knowledge which follows’. this means that inferential knowledge is necessarily a knowledge which is to be preceded by some other knowledge. in other words, anumana consists of two stages, one pertaining to the preceding and the other to the succeeding knowledge. but the two cases of knowledge must have a particular type of relationship known as linga-lingi-bhava which implies that the succeeding one should necessarily come from the preceding. the preceding knowledge has to be in the form of linga. a linga is defined as that which is a necessary mark of something other than itself.3
‘lingin’ stands for that which is marked by linga. between linga and the lingin there is always a gamya-gamaka-bhava which can roughly be regarded as the relation of entailment such that every case of the presence of linga is necessarily a case of the presence of lingin and every case of absence of lingin is the case of the absence of linga.4
in the buddhist tradition linga and lingin are in the form of concepts (vikalpas) and not objects or events or meta-physical reals as they are taken in the nyaya tradition. between any two concepts there will be gamya-gamaka-bhava if and only if they have avinabhava i.e., necessary connection. it is the presence of the necessary connection which is the basis for the passage from the one to the other. this relationship of avinabhava is also known as vyapti. vyapti therefore constitutes the very basis of the inferential process.5
nyaya analysis of anumana: in order to have a better understanding of the buddhist analysis of ‘anumana’ it will be worthwhile if we discuss in brief the nyaya analysis of ‘anumana’. according to nyaya, anumana is the knowledge of an object on the basis of the cognition of its mark along with a remembrance of a previous knowledge concerning an invariable and unconditional relation between the object and its mark.
in other words, in every case of anumana in the preceding cognition, which can be treated as a premise, there are two elements, viz., (i) perceptual cognition of the linga (paksadharmata), and
(ii) the remembrance of unconditional and invariable relation between the linga and the lingin (vyapti). the perceptual cognition of the mark leads to the remembrance of its unconditional and invariable relationship with the lingin resulting in a synthesized knowledge. the synthesis of both these stages is named as paramarsa, which is therefore defined as ‘vyapti visista paksadhamatajnanam’. the act of paramarsa can thus be said to consist of three elements, viz.,
the knowledge of vyapti,
the knowledge of paksadharmata and
the knowledge of the vyapti qualifying paksadharmata.
it is only this unification of paksadharmata and vyapti in paramarsa which entails inferential knowledge. thus though vyapti is one of the causal conditions, and a necessary causal condition (karana), yet it is not the sufficient condition of inference. the sufficient condition (vyapara) is paramarsa only.
buddhist rejection of paramarsa: the buddhist logicians do not draw a sharp distinction between the paksadharmata and vyapti in the way in which the nyaya logicians do. according to them, paksadharmata and vyapti are both comprehended under the concept of trairupya linga and therefore there is no point in talking of vyapti qualifying paksadharmata. thus the nyaya notion of paramarsa is not acceptable to the buddhists.
constituents of inference: in the buddhist analysis, the process of inference involves three basic terms and their interrelations. this is quite evident from the following definition of inference given by dharmakirti, “trirupallingadanumeye yajjnanam tatsvarthanumanam”.6 the three terms are paksa (anumeya), hetu (linga) and sadhya (lingin).
paksa: paksa is the subject under consideration in the inferential reasoning. every inference pertains to some individual or class of individuals about which we want to prove something. hence paksa is that individual or class of individuals about which we want to establish something. it is also named as anumeya because it is the object about which something is to be inferred. in a special sense it also means the underlying substratum (dharmin) to which sadhya is to be ascribed as a property. that is why dharmakirti defines anumeya as “jijnasitavisesa sadhya dharmi.”7
hetu: the other term involved in the process of inference is linga or hetu. in fact it is the pivotal element in the process of anumana. it is the necessary mark which leads to the inference of its marked object. hetu (linga) has three formal characteristics—the satisfaction of which alone enables it to act as a sufficient reason for the inference of its marked object. a hetu which possesses these three characteristics is known as sadhetu.
sadhya: the third entity involved in the inferential process is sadhya or lingin. it is this which constitutes the property (dharma) which is to be inferred in relation to the paksa.
sapaksa: another significant concept which is given in the analysis of the inferential process is sapaksa. sapaksa means an object similar to paksa. in other words, all those objects which possess the property which is to be inferred are known as sapaksa; for example, if fire is the predicate which is to be inferred in relation to a hill, then all those instances like kitchen etc., where fire is known to be a predicate, constitute sapaksa. a sapaksa is similar to paksa in this sense only that both of them comprehend a common property.8
asapaksa: a case which is not similar to paksa is regarded as asapaksa.9 in other words, asapaksa is that which is never a possessor of the property commonly possessed by paksa and sapaksa. asapaksa can be of three types:10
(a) different from it (anya).
(b) contrary to it (viruddha).
(c) absence of it (abhava).
vyapti: the entire inferential process, as we have said above, is based upon the relation between linga and lingin, which can be understood in terms of necessary dependence (avinabhavaniyama) and which is technically known as vyapti. the buddhist conception of vyapti stands for an invariable necessary connection. vyapti is a necessary bond because of the fact that it is rooted in what is technically known as svabhava pratibandha or existential dependence. existential dependence means dependent existence. it may be in the form of a causal relation or an analytical entailment. for example, the dependence of effect on its cause enables us to infer the cause the moment the effect is known to us. similarly, an analytically deduced fact by its very essence depends upon the fact from which it is deduced. thus there is svabhava pratibandha between cause and effect and between the deduced object and that from which there is deduction. the example of the former type is the relation between smoke and fire and of the latter type is the relation between rose and flower. we can deduce one fact from another only if there is existential dependence. it can be asked why is it that we can deduce one fact from another only if there is existential dependence.11 the answer given by the buddhist logicians is that this is so because effect which is not dependent upon another object cannot be invariably and necessarily concomitant with the later. in other words, if effect is not tied up by its existence to another object, it can not be necessarily concomitant with the latter. there will be no invariability (avyabhicara). thus the possibility of deducing one fact from the other depends upon an invariable and necessary connection which precludes the existence of the one without the existence of the other. therefore, if two facts are existentially connected we can assert that one of them can not exist independently of the other and therefore from the presence of the one follows the presence of the other.
kinds of linga: there are three varieties of linga, viz., anupalabdhi, svabhava and karya. the lingin is a sort of predicate and a predicate is either denied or affirmed. when it is denied, this is done on the basis of the non-existence of its mark. such a mark is known as anupalabdhi hetu or anupalabdhi linga. when it is affirmed, its mark is either existentially identical with it or if different, it is its effect. in the former case its linga is known as svabhava hetu or svabhava linga and in the latter case it is known as karya hetu or karya linga.
anupalabdhi: anupalabdhi has been defined as non-cognition of such an object which otherwise fulfils the conditions of cognizability. for example, a jar is an object which fulfils the condition of cognizability. if at a particular place there is non-cognition of jar, this enables us to infer its non-existence. so here non-cognition of the jar is the linga and non-existence of the jar is the lingin. the non-cognition (of a thing) is to be regarded as the linga for the non-existence (of that thing) which is its lingin on the ground that if the things were present, it would have necessarily been perceived when all other conditions of perceptibility are fulfilled. if inspite of all the conditions of perceptibility being present, if a thing is not perceived, we can legitimately infer its non-existence.12
svabhava linga or svabhava hetu: the second type of linga is known as svabhava linga or svabhava hetu. the svabhava hetu is defined as that whose mere existence is sufficient for the deduction of sadhya.13 for example, in the judgement, “it is a flower because it is a rose” the reason, namely, rose is sufficient for the deduction of flower. here the terms ‘rose’ and ‘flower’ have one and the same object for their reference though they may have different meanings. it is this sameness of reference known as tadatmya which is responsible for the existential tie between rose and flower.
karya linga or karya hetu: the third type of linga is karya linga or karya hetu, which is in the form of an effect. it necessarily presupposes its cause like smoke necessarily implying the existence of fire. the causal connection is given to us in our experience of both anvaya and vyatireka type i.e., on the basis of agreement in presence and agreement in absence between two phenomena.14
three types of anumana: since there are three types of linga, there are three types of anumana, viz., anupalabdhi, svabhava and karya.
though a linga may be either in the form of anupalabdhi or svabhava or karya, every linga necessarily possesses three marks. the doctrine of three marks of a linga, technically known as trairupyavada, is of great logical significance in the buddhist theory of inference.
the concept of trairupya: the concept of linga provides the starting point of the inferential process. if the two stages of the inferential process are classified as premise and conclusion then linga can be regarded as the most basic concept in the premise. the process of inference consists of a transition from linga to lingin. that is why dharmakirti regards linga as the very basis of inference.15 regarding the function of linga dharmottara aptly remarks, “therefore the function of the logical mark, owing to which it is able to create cognition of not directly known things, is nothing else than the necessity of an invariable concomitance between the perceived mark and the non-perceived object. it follows that the world ‘necessary’ must be referred to all the three aspects in which the mark manifests itself, since all these three viz. (i) the positive concomitance of the mark with the deduced predicate; (ii) its contraposition (or the inverted) concomitance of their negations) and (iii) the presence of the thus characterized mark upon the subject of the conclusion represent the essence of the function performed by a logical mark and must be ascertained as being necessary”.16
in this passage dharmottara refers to a very significant term, viz., nantariyakatva which has also been referred to by uddyotakara, a nyaya thinker in his nyaya-vartika. he has referred to a definition of anumana given by some thinkers as “nantariyakartha darsanam tadvido anumanam” i.e. “the experience of a thing, which is inseparably connected with another thing is the instrument of inference for one who knows that they are inseparably connected.”17 the concept of nantariyakatva is perhaps explicated by dignaga in the form of trairupya. the term nantariyakartha implies presence of hetu in the paksa, its presence in the sapaksa and its absence in the vipaksa, because in the absence of such a situation there can not be inseparable connection between hetu and sadhya.
according to buddhist logic the linga is, thus, characterized by three essential characteristics. in fact in the history of indian logic we find different views with regard to the essential characteristics of linga. whereas the nyaya tradition insists on five characteristics and the jaina tradition regards only one characteristic, the buddhist tradition maintains that there are three and only three essential characteristics of a linga. every linga must possess all the three characteristics simultaneously (trilaksana hetu). then and then only it can be regarded as a linga, and be made use of in the process of inference. ( that is why dharmakirti, while defining anumana, writes trirupallingad etc.
references
1. (a) savyapara-pratitatvat pramanam phalamevasat, p.s. i.8.
(b) visayakarataivasya pramanam tena miyate. p.s. i.9.
2. see p.s. v. i.8 hattori’s translation, p.28.
3. lingyate gamyate iti lingah n.b.t. p.21.
4. niscayapekso hi gamya-gamaka bhavah n.b.t. p.31.
5. na hi yo yatra svabhavena na pratibaddhah sa tamaprati baddha-visayamavasyameva na vyabhicaratiti nasti tayora vyabhicaraniyamah. avinabhavaniyamab. avyabhicara niyamacca gamya-gamaka-bhavah. n.b.t., p. 30.
6. n.b. ch. ii, p. 21.
7. ibid., p.24.
8, (a) sadhya-dharma-samanyena samano’rthah sapaksah, n.b. ch. ii. p.24
(b) samanah sadrso yo’rthah paksena sa sapaksa…n.b.t., p.24
9. na sapakso’ sapaksah…..n.b. ch.ii. p.215.
10. tato’nyastadviruddhastadabhavasceti, ibid.
11. svabhava-pratibandhe hi satyartho’itham gamayeta, n.b. ch.ii. p.29.
12. upalabdhi laksana praptasyanupalabdheriti, n.b. ch.ii. p.26.
13. svabhavah svasattamatrabhavini sadhya-dharme hetuh. n.b. ch. ii. p.28.
14. karya-karana-bhavo loke pratyaksanupalambhanibandhanah pratiti. n.b.t. p.28.
15. triupallingat etc. n.b., ch. ii, p.21.
16. tasmat paroksartha nantariyakataya niscayanamevalingasya paroksartha pratipadanavyaparah. naparah kascita. ato’ nvaya vyatireka paksa dharmatva niscayo lingavyaparamakatvadavasya karttavya iti sarvesu rupesu niscita grahanamapeksaniyam.
17. n.v. 1.1.5.
the buddhist theory of inference
s.r. bhatt
the epistemological thinkers in india have generally adopted a causal approach to knowledge. knowledge is taken to be an occurrence, an outcome of a particular causal complex (karana samagri) in which the causal condition acting as an instrumental cause (karana) is known as pramana. pramana is the mode of knowing. the buddhist thinkers do not entertain the distinction between pramana and its outcome (pramana phala= prama) mainly because this distinction is not needed in their epistemological set up. as opposed to the school of nyaya which maintains such a distinction because of its presupposition that pramana is the ground for the truth of a prama (manadhina meyasiddhih) which is its phala (outcome), the buddhists repudiate this distinction because for them a reference to the object of knowledge (prameya) is the ground for the truth of prama.1 moreover, they maintain that no rigid separation is possible between the act of cognizing and the cognition of the object.2
anumana, which in the buddhist tradition is one of the two pramanas, is at once a mode of knowing and a way of reasoning. thus it has an epistemic as well as a logical aspect.
the word anumana literally means ‘a knowledge which follows’. this means that inferential knowledge is necessarily a knowledge which is to be preceded by some other knowledge. in other words, anumana consists of two stages, one pertaining to the preceding and the other to the succeeding knowledge. but the two cases of knowledge must have a particular type of relationship known as linga-lingi-bhava which implies that the succeeding one should necessarily come from the preceding. the preceding knowledge has to be in the form of linga. a linga is defined as that which is a necessary mark of something other than itself.3
‘lingin’ stands for that which is marked by linga. between linga and the lingin there is always a gamya-gamaka-bhava which can roughly be regarded as the relation of entailment such that every case of the presence of linga is necessarily a case of the presence of lingin and every case of absence of lingin is the case of the absence of linga.4
in the buddhist tradition linga and lingin are in the form of concepts (vikalpas) and not objects or events or meta-physical reals as they are taken in the nyaya tradition. between any two concepts there will be gamya-gamaka-bhava if and only if they have avinabhava i.e., necessary connection. it is the presence of the necessary connection which is the basis for the passage from the one to the other. this relationship of avinabhava is also known as vyapti. vyapti therefore constitutes the very basis of the inferential process.5
nyaya analysis of anumana: in order to have a better understanding of the buddhist analysis of ‘anumana’ it will be worthwhile if we discuss in brief the nyaya analysis of ‘anumana’. according to nyaya, anumana is the knowledge of an object on the basis of the cognition of its mark along with a remembrance of a previous knowledge concerning an invariable and unconditional relation between the object and its mark. in other words, in every case of anumana in the preceding cognition, which can be treated as a premise, there are two elements, viz., (i) perceptual cognition of the linga (paksadharmata), and (ii) the remembrance of unconditional and invariable relation between the linga and the lingin (vyapti). the perceptual cognition of the mark leads to the remembrance of its unconditional and invariable relationship with the lingin resulting in a synthesized knowledge. the synthesis of both these stages is named as paramarsa, which is therefore defined as ‘vyapti visista paksadhamatajnanam’. the act of paramarsa can thus be said to consist of three elements, viz., the knowledge of vyapti, the knowledge of paksadharmata and the knowledge of the vyapti qualifying paksadharmata. it is only this unification of paksadharmata and vyapti in paramarsa which entails inferential knowledge. thus though vyapti is one of the causal conditions, and a necessary causal condition (karana), yet it is not the sufficient condition of inference. the sufficient condition (vyapara) is paramarsa only.
buddhist rejection of paramarsa: the buddhist logicians do not draw a sharp distinction between the paksadharmata and vyapti in the way in which the nyaya logicians do. according to them, paksadharmata and vyapti are both comprehended under the concept of trairupya linga and therefore there is no point in talking of vyapti qualifying paksadharmata. thus the nyaya notion of paramarsa is not acceptable to the buddhists.
constituents of inference: in the buddhist analysis, the process of inference involves three basic terms and their interrelations. this is quite evident from the following definition of inference given by dharmakirti, “trirupallingadanumeye yajjnanam tatsvarthanumanam”.6 the three terms are paksa (anumeya), hetu (linga) and sadhya (lingin).
paksa: paksa is the subject under consideration in the inferential reasoning. every inference pertains to some individual or class of individuals about which we want to prove something. hence paksa is that individual or class of individuals about which we want to establish something. it is also named as anumeya because it is the object about which something is to be inferred. in a special sense it also means the underlying substratum (dharmin) to which sadhya is to be ascribed as a property. that is why dharmakirti defines anumeya as “jijnasitavisesa sadhya dharmi.”7
hetu: the other term involved in the process of inference is linga or hetu. in fact it is the pivotal element in the process of anumana. it is the necessary mark which leads to the inference of its marked object. hetu (linga) has three formal characteristics—the satisfaction of which alone enables it to act as a sufficient reason for the inference of its marked object. a hetu which possesses these three characteristics is known as sadhetu.
sadhya: the third entity involved in the inferential process is sadhya or lingin. it is this which constitutes the property (dharma) which is to be inferred in relation to the paksa.
sapaksa: another significant concept which is given in the analysis of the inferential process is sapaksa. sapaksa means an object similar to paksa. in other words, all those objects which possess the property which is to be inferred are known as sapaksa; for example, if fire is the predicate which is to be inferred in relation to a hill, then all those instances like kitchen etc., where fire is known to be a predicate, constitute sapaksa. a sapaksa is similar to paksa in this sense only that both of them comprehend a common property.8
asapaksa: a case which is not similar to paksa is regarded as asapaksa.9 in other words, asapaksa is that which is never a possessor of the property commonly possessed by paksa and sapaksa. asapaksa can be of three types:10
(d) different from it (anya).
(e) contrary to it (viruddha).
(f) absence of it (abhava).
vyapti: the entire inferential process, as we have said above, is based upon the relation between linga and lingin, which can be understood in terms of necessary dependence (avinabhavaniyama) and which is technically known as vyapti. the buddhist conception of vyapti stands for an invariable necessary connection. vyapti is a necessary bond because of the fact that it is rooted in what is technically known as svabhava pratibandha or existential dependence. existential dependence means dependent existence. it may be in the form of a causal relation or an analytical entailment. for example, the dependence of effect on its cause enables us to infer the cause the moment the effect is known to us. similarly, an analytically deduced fact by its very essence depends upon the fact from which it is deduced. thus there is svabhava pratibandha between cause and effect and between the deduced object and that from which there is deduction. the example of the former type is the relation between smoke and fire and of the latter type is the relation between rose and flower. we can deduce one fact from another only if there is existential dependence. it can be asked why is it that we can deduce one fact from another only if there is existential dependence.11 the answer given by the buddhist logicians is that this is so because effect which is not dependent upon another object cannot be invariably and necessarily concomitant with the later. in other words, if effect is not tied up by its existence to another object, it can not be necessarily concomitant with the latter. there will be no invariability (avyabhicara). thus the possibility of deducing one fact from the other depends upon an invariable and necessary connection which precludes the existence of the one without the existence of the other. therefore, if two facts are existentially connected we can assert that one of them can not exist independently of the other and therefore from the presence of the one follows the presence of the other.
kinds of linga: there are three varieties of linga, viz., anupalabdhi, svabhava and karya. the lingin is a sort of predicate and a predicate is either denied or affirmed. when it is denied, this is done on the basis of the non-existence of its mark. such a mark is known as anupalabdhi hetu or anupalabdhi linga. when it is affirmed, its mark is either existentially identical with it or if different, it is its effect. in the former case its linga is known as svabhava hetu or svabhava linga and in the latter case it is known as karya hetu or karya linga.
anupalabdhi: anupalabdhi has been defined as non-cognition of such an object which otherwise fulfils the conditions of cognizability. for example, a jar is an object which fulfils the condition of cognizability. if at a particular place there is non-cognition of jar, this enables us to infer its non-existence. so here non-cognition of the jar is the linga and non-existence of the jar is the lingin. the non-cognition (of a thing) is to be regarded as the linga for the non-existence (of that thing) which is its lingin on the ground that if the things were present, it would have necessarily been perceived when all other conditions of perceptibility are fulfilled. if inspite of all the conditions of perceptibility being present, if a thing is not perceived, we can legitimately infer its non-existence.12
svabhava linga or svabhava hetu: the second type of linga is known as svabhava linga or svabhava hetu. the svabhava hetu is defined as that whose mere existence is sufficient for the deduction of sadhya.13 for example, in the judgement, “it is a flower because it is a rose” the reason, namely, rose is sufficient for the deduction of flower. here the terms ‘rose’ and ‘flower’ have one and the same object for their reference though they may have different meanings. it is this sameness of reference known as tadatmya which is responsible for the existential tie between rose and flower.
karya linga or karya hetu: the third type of linga is karya linga or karya hetu, which is in the form of an effect. it necessarily presupposes its cause like smoke necessarily implying the existence of fire. the causal connection is given to us in our experience of both anvaya and vyatireka type i.e., on the basis of agreement in presence and agreement in absence between two phenomena.14
three types of anumana: since there are three types of linga, there are three types of anumana, viz., anupalabdhi, svabhava and karya.
though a linga may be either in the form of anupalabdhi or svabhava or karya, every linga necessarily possesses three marks. the doctrine of three marks of a linga, technically known as trairupyavada, is of great logical significance in the buddhist theory of inference.
the concept of trairupya: the concept of linga provides the starting point of the inferential process. if the two stages of the inferential process are classified as premise and conclusion then linga can be regarded as the most basic concept in the premise. the process of inference consists of a transition from linga to lingin. that is why dharmakirti regards linga as the very basis of inference.15 regarding the function of linga dharmottara aptly remarks, “therefore the function of the logical mark, owing to which it is able to create cognition of not directly known things, is nothing else than the necessity of an invariable concomitance between the perceived mark and the non-perceived object. it follows that the world ‘necessary’ must be referred to all the three aspects in which the mark manifests itself, since all these three viz. (i) the positive concomitance of the mark with the deduced predicate; (ii) its contraposition (or the inverted) concomitance of their negations) and (iii) the presence of the thus characterized mark upon the subject of the conclusion represent the essence of the function performed by a logical mark and must be ascertained as being necessary”.16
in this passage dharmottara refers to a very significant term, viz., nantariyakatva which has also been referred to by uddyotakara, a nyaya thinker in his nyaya-vartika. he has referred to a definition of anumana given by some thinkers as “nantariyakartha darsanam tadvido anumanam” i.e. “the experience of a thing, which is inseparably connected with another thing is the instrument of inference for one who knows that they are inseparably connected.”17 the concept of nantariyakatva is perhaps explicated by dignaga in the form of trairupya. the term nantariyakartha implies presence of hetu in the paksa, its presence in the sapaksa and its absence in the vipaksa, because in the absence of such a situation there can not be inseparable connection between hetu and sadhya.
according to buddhist logic the linga is, thus, characterized by three essential characteristics. in fact in the history of indian logic we find different views with regard to the essential characteristics of linga. whereas the nyaya tradition insists on five characteristics and the jaina tradition regards only one characteristic, the buddhist tradition maintains that there are three and only three essential characteristics of a linga. every linga must possess all the three characteristics simultaneously (trilaksana hetu). then and then only it can be regarded as a linga, and be made use of in the process of inference. ( that is why dharmakirti, while defining anumana, writes trirupallingad etc.
development of the concept of trairupya: the first systematic formulation of the doctrine of trairupya, is said to be done by dignaga. uddyotakara, in his vartika cites dignaga’s formulation of trairupya as follows:
“anumeya’tha tat-tulye sadbhavo nastita sati”18 i.e. existence in the anumeya (paksa), in what is like the paksa (sapaksa) and absence in what is not like the paksa (vipaksa). prof stcherbatsky has rendered in english dignaga’s version as follows:19
1. its presence in the subject of the inference.
2. its presence in similar instances.
3. its absence in the dissimilar instances.
it was dhamakirti who further regulated these three characteristics introducing more rigour.
he did so by putting and emphasising the word ‘eva’ to each of the three marks, and by qualifying the entire expression by the word niscitam. the other modification he introduced is about the use of the terms sapaksa and asapaksa to remove all sorts of ambiguity. thus his formulation is as follows:20
1. lingasyanumeye sattvameva.
2. sapaksa eva sattvam.
3. asapakse casattvameva niscitam.
in english it can be rendered as follows:
1. existence only (never non-existence) in the paksa.
2. existence in all those things which are similar to paksa (never in things which are not similar to paksa).
3. only non-existence (never existence) in things which are not similar to paksa.
professor stcherbatsky21 has rendered dharmakirti’s version in three stages as follows:
i stage:
(i) the presence of the reason in the subject, its presence ‘just’ i.e., never absence.
(ii) its presence in similar instances, ‘just’ in similars, i.e., never in dissimilars, but not in the totality of similars.
(iii) its absence from dissimilar instances, its absence ‘just’ i.e., never presence, absence from the totality of the dissimilar instances.
it is easily seen that the second and the third rule mutually imply each other. if the reason is present in the similar instances only, it also is absent from every dissimilar case. and if it is absent from every dissimilar case, it can be present in similar instances only, although not necessarily in all of them. nevertheless both the rules must be mentioned, because, although in a correct inference the application of the one means the application of the other, in a logical fallacy their infringements sometimes carry different results. dharmakirti, moreover, adds the word ‘necessary’ to the formulation of each rule. their final form will thus be:
ii stage:
(i) the necessary presence of the reason in the subject’s totality.
(ii) its necessary presence in similars only, although not in their totality.
(iii) its necessary absence from dissimilars in their totality.
iii stage:
the third and the most succinct formulation as follows:
(i) in subject wholly.
(ii) in similar only.
(iii) in dissimilar never.
though the doctrine of trairupya has been one of the most significant elements in buddhist logic, the formulation of it both by dignaga and by dharmakirti does not seem to be free from ambiguity. that is why the interpretation of ‘trairupya’ has always been controversial. prof. chi is to the point when he remarks that, “the trairupya seems to have been respected as one of the most important doctrines in indian logic, and yet elaborate interpretations of distinguished scholars can hardly convey any sense to their readers.”22
the discussions recorded in surviving documents are lengthy. some were made by commentators who tried hard to give plausible interpretations without caring very much what dignaga should have meant to say; some others were made by dignaga’s opponents such as uddyotakara, who might have twisted his words in order to attack him.
uddyotakara’s objection to dignaga’s formulation: the formulation of the first rupa in dignaga viz., ‘anumeye sadbhava’ was not free from ambiguity and as a consequence of that uddyotakara23 examined its two possible interpretations viz., (i) the hetu is present in the subject only, (2) the hetu is only present in the subject.
to point out the difference in the two interpretations the following symbolic formulation may be helpful.
(i) b is present in a only = every b is a = no non-a is b.
(ii) only b is present in a = every a is b = no non-b is a.
from the above it is quite evident that these two interpretations do not mean the same thing
and therefore, the formulation by dignaga was replete with ambiguity.
the formulation of the second rupa given by dignaga was also ambiguous in the same way which led uddyotakara to interpret it in two different ways, viz.,
(iia) the hetu is present in similar instances only.
(iib) the hetu is present in ‘all similar instances’.
in the same way dignaga’s formulation of the third rupa was subjected to two different interpretations by uddyotakara as follows:
(iiia) only the hetu is absent in the ‘dissimilar instances’.
(iiib) the hetu is absent in the dissimilar instances only.
though dignaga had given his unambiguous version in his ‘hetu cakra damaru’ where the presence or the absence of hetu in relation to anumeya was well quantified, perhaps he did not do so in the pramana-samuccaya, from where uddyotakara might have quoted him.
dharmakirti’s interpretation: the ambiguity of non-quantification which vitiated dignaga’s formulation was sought to be removed by dharmakirti by adding the restrictive word ‘eva’. but unexpectedly his modification caused further ambiguity. that is why commentators from dharmottara onwards had to struggle hard to interpret it in a satisfactory way. dharmottara had to face a dilemma on this account while commenting upon the second and third rupas. the second rupa apparently meant only in the sapaksas hetu must be present but by implication this meant that hetu must be absent in all vipaksa. but then second rupa will be another way of saying the same thing as the third rupa. this renders the third rupa superfluous but this was not the intention either of dignaga or of dharmakirti.
points of ambiguity in the formulation of trairupya: there are three main reasons for the misinterpretation of the theory of trairupya. the first reason for the ambiguity was the use of terms like ‘anumeya’ which may either mean ‘sadhyadharma’ or ‘sadhyadharmin’. a similar ambiguity was present in the use of the term ‘eva’ which was introduced as a restrictive predicate. in the first rupa, as it is formulated by dharmakirti, the word ‘eva’ could be applied as a restriction to the occurrence of the word ‘anumeya’ or of the word ‘linga’ or of the word ‘sattvam’. in the formulation of the second rupa the word ‘eva’ can be understood as restriction of either sapaksa or linga or sattvam. similarly in the formulation of the third rupa ‘eva’ can be a restriction either for asapaksa or for linga or for a sattvam. it was dharmottara who later on tried to remove this ambiguity of placement of ‘eva’.[j1]
the second reason for the ambiguity in the formulation of trairupya is non-availability of dignaga’s works in their complete form. from the fragments what one gets is only disconnected and isolated thoughts and not the complete theory.
the third reson for the ambiguity has been the twist given by uddyotakara to dignaga’s position.
chinese formulation of trairupya: prof. chi in his book ‘buddhist formal logic’ has given the chinese version of trairupya which seems to be free from ambiguity.
it is as follows:24
(i) the ‘pervasive presence’ of the hetu in the subject.
(ii) the ‘necessary presence’ of the hetu in some similar instances.
(iii) the ‘pervasive absence’ of the hetu from dissimilar instances.
in the above rendering the use of the word ‘pervasive’ in the first and the third rupas and of the word ‘necessary’ in the second rupa helps us in the removal of ambiguity. the words ‘pervasive presence’ and ‘pervasive absence’ refer to the distribution of quantification of hetu. similarly the phrase ‘necessary-presence’ means ‘assured presence’ or ‘not failing to be present’. prof. chi has defined the notion of ‘pervasive presence’ as follows:
“pervasive presence of b in a”
= “b is present in every a ”
= “every a is b”
the notion of ‘pervasive absence’ is defined by him as follows:
“pervasive absence of b from a “
=”b is absent from every a”
=”every a is non-b”
=”no a is b”
the notion of ‘necessary presence’ is defined by him as follows:
“necessary presence of b in a”
=”b is present in at least one a, at most in every a”
=”at least one a, at most every a is b”.
in accordance with the above understanding he interprets trairupya as follows:
ist rupa: “the property g is present in everything which possesses the property f “
=”everything which possesses the property f possesses the property g”
= for every x, x is an f imples x is g”.
2nd rupa: “there is at least one occasion in which the property g is present in a thing which
possesses the property h, apart from the thing which possesses the property f, which
remains to be proved”.
= apart from the thing which possesses the property f, at least one thing which
possesses the property h possesses the property g”.
=”for some x which is not f, x is both an h and g”.
3rd rupa: “there is no occasion in which the property g is present in things which possess the
property of non-h.”
= “nothing which possesses the property non-h possesses the property g”.
= “for no x, x is both non-h and a g.” on the basis of this interpretation prof. chi
proposes to resolve dharmakirti’s dilemma on the interpretation of ‘eva’ only in the
second rupa as follows:
‘only presence’
“only presence of the property of the hetu in similar instances but not otherwise”
=”only presence of the property of the hetu in similar instances but utter absence of
the property of it in all similar instances”.
=”the presence of the property of the hetu in at lest one similar instance”.
understood in this way the second rupa does not overlap with the third rupa and the problem of redundancy is overcome.
‘trairupya’ and ‘hetu cakra’: the doctrine of trairupya can unambiguously be understood only in the context of hetucakra, another doctrine propounded by dignaga in his small tract called ‘hetucakra damaru’. the application of ‘hetucakra’ to the second and third rupa gives the following which is represented by prof. chi in a diagram as under:
hetucakra trairupya
sapaksavyapaka(1)
sapaksa-evasattvam(5)
vipaksavrtti(2)
sapaksaikadesavrtti(3)
asapakse casattvameva (6)
vipaksavrtti(4)
the above diagram can be explicated in ordinary language as follows:
(1) pervasive presence in similar instances.
(2) absence in dissimilar instances.
(3) partial presence in similar instances.
(4) absence in dissimilar instances.
(5) unfailing presence in similar instances.
(6) necessary absence in dissimilar instances.
there are three significant systems of logic in india, viz., buddhist, nyaya and jaina.
all the three systems have differed with regard to the required rupas of hetu. the buddhists, as we have seen, insist on three rupas only, but the naiyayikas emphasise upon five rupas of hetu and the jainas maintain that only one rupa of hetu is necessary.
nvaya theory of pancarupya: the nyaya logicians emphasise the fact that hetu or linga
is the central element in the inferential process. it is related to paksa as well as sadhya and it occurs in all the significant premises, viz., hetu, udaharana and upanaya. the validity of inference depends to a great extent on hetu and that is why a fallacious inference is known as having hetvabhasa. the nyaya logicians, therefore, have taken great pain in expounding the characteristics of a good hetu, because if the hetu is sadhetu, the inference will be valid and if it is asadhetu, the inference will be invalid. the naiyayikas enumerate the following conditions of a sadhetu.
(1) paksadharmatvam: hetu must be a dharma or characteristic of paksa. this means that
the hetu should have adhikarana in paksa, i.e., it should reside in or be related to paksa. thus the paksa should never be without hetu. and nothing can be a paksa if there is no hetu related to it.
(2) sapaksasattvam: hetu must be distributively related to sadhya. this means that wherever
hetu exists in all such cases, sadhya must be present. this means that the necessary condition for the occurrence of hetu, is the occurrence of sadhya.
(3) vipaksasattvam: hetu must be absent in all cases in which the sadhya is absent. this
means that the absence of sadhya necessarily implies the absence of the hetu. hetu cannot be present in the absence of sadhya. so in all vipaksas i.e. where sadhya is absent hetu also must be absent.
(4) abadhitavisayatvam: hetu should not be such as to be contradictory to or incompatible
with the nature of sadhya. this is so because the function of hetu is to establish sadhya, but if it is incompatible with sadhya, it will fail to do so.
(5) asatapratipaksatvam or aviruddhatvam: hetu should not be such which can have a rival
hetu which is competent to invalidate it.
of course naiyayikas make it clear that all these five rupas are not simultaneously
present in all hetus. for example, in a kevalanvayi inference the third condition will not be applicable. similarily in the kevalavyatireki inference, the second condition will not be applicable to its hetu. hence it has been said that a sadhetu is one that satisfies the five or at least the four conditions as explained above.
if we compare the nyaya theory of panca rupa hetu with the buddhist theory of trairupya, we find that the first three rupas of the naiyayikas correspond to the trairupya of the buddhists. of course the details in the formulation of these three rupas are very much different in the two traditions, still we can regard them as having some rough correspondence. but the last two rupas of the nyaya theory are not accepted and will not be acceptable to the buddhists. now it can be asked as to why the buddhists do not accept the last two rupas? the reason is that the buddhists believe in the theory of pramanavyavastha which is opposite to the nyaya theory of pramana-samplava. so verdict of one paramana can be contradicted by the other pramana in the nyaya position but it can not be so in the buddhist position. therefore there is no possibility of satpratipaksa in the buddhist position. secondly, the basis of inference according to the buddhists is linga-lingi-bhava between any two concepts. we shall never establish linga-lingi-bhava if they are of opposite nature. therefore, there will never be an occasion for abadhitavisayatvam. though there has been a lot of controversy between the buddhists and the naiyayikas as to whether only three rupas are to be accepted, or five rupas are to be accepted, the whole controversy appears to be futile because they are two different traditions having different types of requirements. the nyaya tradition stands in need of five rupas, whereas in the buddhist tradition only three rupas are possible and therefore, the theories of pancarupya hetu and trairupya hetu are perfectly in order in their respective systems.
jaina criticism of trairupya: the jaina tradition maintains that hetu has a unitary nature and therefore, it possesses only one rupa. hetu is defined as that which is inseparably connected with sadhya. this means hetu is that which is present only in the presence of sadhya and which is necessarily absent in the absence of sadhya. akalankadeva defines hetu as “sadhyavinabhavavinibodhaikalaksana”.25 some jaina logicianshave described this fact with the help of another concept, viz. “anyathanupapatti”. vadidevasuri defines hetu as “niscitanyathanupapattyeka laksanohetuh.”26 the term “anyathanupapatti” means that in the absence of sadhya (anyatha) there should be absence of hetu (anupapatti). by implication it means that hetu should be present only in the presence of sadhya. so to be sadhyavinabhavini is the sole characteristic of a hetu.
the jainas have not only insisted on one and the only one characteristic of hetu but they have also put forth a refutation of the theories of ‘trairupya’ and ‘pancarupya’.
patrasvamin seems to be the first jaina scholar, as quoted by akalanka and vidyananda who attempted the refutation of the doctrine of trairupya as follows:27
“anyathanupapannatvam yatra tatra trayena kim,
nanyathanupapannatvam yatra tatra trayena kim.”
in the above karika patrasvamin emphatically maintains that ‘anyathanupapannatvam’ is the only rupa of hetu. if it is present then what is the need of three rupas, and if it is absent then what is the purpose of having three rupas. following patrasvamin all subsequent jaina logicians like akalanka, vidyananda, prabhacandra, vadidevasuri, anantavirya, hemacandra etc. have also refuted the doctrine of trairupya. the above mentioned karika of patrasvamin has been quoted by santaraksita in his tattvasangraha in the form of purva paksa presenting the jaina position and there he has refuted the jaina objections against ‘trairupya’.28 the basic objection put forth by the jainas is that even an invalid hetu may have the three rupas. for example, the argument, “there is smoke on the hill because there is fire” has all the three rupas, viz. paksadharmattvam, sapaksa-sattvam and vipaksa-sattvam, but still it is logically invalid because there is no necessary connection (avinabhava) between fire and smoke. in other words, there may be fire and smoke on the hill and so the condition of paksasattvam is fulfilled. similarly there can be fire and smoke in the kitchen and condition of sapaksa-sattvam is fulfilled. likewise there will be no fire and smoke in a lake and the condition of vipaksasattvam is also fulfilled. even then fire is an invalid hetu for smoke. of course, the buddhists may point out that the trairupya should be understood as qualified by avinabhava and there is no avinabhava between fire and smoke because there may be fire without smoke, for example, in an electrical hearth. so the relation of fire and smoke is conditional (sopadhika) and not unconditional (nirupadhika). to this jainas have replied that if avinabhava is to be emphasised, then this amounts to accepting their position, which means hetu has only one rupa and not three rupas. the jaina point of emphasis is that the hetu in an argument is valid and competent to prove the sadhya not because the two exist in the same paksa as the naiyayikas say but because the two are inseparably related. in the words of jaina logicians the hetu is not cognised otherwise than in connection with the sadhya. accordingly, so far as the essential characteristics of a valid hetu are concerned, it is this one and this one only.
santaraksita in his tattvasangraha has stated and examined patrasvamin’s arguments in great detail. according to him the concept of anyathanupapatti is only an abbreviated form of trairupya, the like of which has also been given by dignaga as ‘hetordharminyevagamyate’ 29 )
references
8. (a) savyapara-pratitatvat pramanam phalamevasat, p.s. i.8.
(b) visayakarataivasya pramanam tena miyate. p.s. i.9.
9. see p.s. v. i.8 hattori’s translation, p.28.
10. lingyate gamyate iti lingah n.b.t. p.21.
11. niscayapekso hi gamya-gamaka bhavah n.b.t. p.31.
12. na hi yo yatra svabhavena na pratibaddhah sa tamaprati baddha-visayamavasyameva na vyabhicaratiti nasti tayora vyabhicaraniyamah. avinabhavaniyamab. avyabhicara niyamacca gamya-gamaka-bhavah. n.b.t., p. 30.
13. n.b. ch. ii, p. 21.
14. ibid., p.24.
8, (a) sadhya-dharma-samanyena samano’rthah sapaksah, n.b. ch. ii. p.24
(b) samanah sadrso yo’rthah paksena sa sapaksa…n.b.t., p.24
9. na sapakso’
