Buddhist Philosophy of Logic
Logic in Buddhist Philosophy concerns the systematic study of anuma¯na (often trans-lated as inference) as developed by Digna¯ga (480-540 c.e.) and Dharmak¯ıti (600-660 c.e.). Buddhist logicians think of inference as an instrument of knowledge (prama¯n. a) and, thus, logic is considered to constitute part of epistemology in the Buddhist tradition. Accord-ing to the prevalent 20th and early 21st century ‘Western’ conception of logic, however, logical study is the formal study of arguments. If we understand the nature of logic to be formal, it is di�cult to see what bearing logic has on knowledge. In this paper, by weaving together the main threads of thought that are salient in Digna¯ga’s and Dharmak¯ıti’s texts, I shall re-conceive the nature of logic in the context of epistemology and demarcate the logical part of epistemology which can be recognised as logic. I shall demonstrate that we can recognise the logical signiﬁcance of inference as understood by Buddhist logicians despite the fact that its logical signiﬁcance lies within the context of knowledge.
Logic in Buddhist Philosophy, as we understand it in this chapter, concerns the systematic study of anuma¯na (often translated as inference) as developed by Digna¯ga (480-540 c.e.) and Dharmak¯ıti (600-660 c.e.). Buddhist logicians think of inference as an instrument of knowledge (prama¯n. a) and, thus, logic is consid-ered to constitute part of epistemology in the Buddhist tradition. The focus of this chapter is on the tradition of Buddhist philosophy called prama¯n. ava¯da which is concerned mainly with epistemology and logic. Thus, we are not concerned with various reasoning patterns that can be discerned in the writings of Bud-dhist philosophers; for example, when the Ma¯dhyamika philosopher Na¯ga¯rjuna presents an argument in terms of catus. ko.ti (four ‘corners’: true, false, both, nei-ther).
2 Nor are we concerned with lists of ‘rules’ for debates, such as the ones contained in the Indian va¯da (debate) literature and Tibetan bsdus grwa, though debates are important aspects of Indo-Tibetan Buddhist intellectual life.3 This chapter contains a discussion of the philosophy of logic that is at-tributable to Buddhist logicians. I will attempt to make sense of the Bud-dhist conception of the nature of logic by weaving together the main treads of thought that are salient in Digna¯ga’s Prama¯n. asamuccaya (and his auto- commentary Prama¯n. asamuccayavr.tti) and Dharmak¯ırti’s Prama¯n. avinis´caya and
Prama¯n. ava¯rttika. The exegetical studies of these texts and other texts recognised as belonging to the prama¯n. ava¯da tradition are extensive but not exhaustive. The collection of historical data and the close reading of these texts are important tasks. In this chapter, however, we step back from the texts and examine (or re-examine) what ‘inference’ or ‘logic’ might mean for Buddhist logicians.4 I will sketch the Buddhist conception of the nature of logic by uncovering some of the presuppositions that underly the thoughts expressed in Buddhist logic texts.
The Role of Inference in Epistemology
Epistemology in the Buddhist philosophical tradition is generally concerned with instruments or sources of knowledge. What we know is considered to depend on how we know. That is, what we are warranted to be aware of depends on how we come to be aware of it. So, in order to understand what we know, Bud-dhist philosophers investigate the sources which give rise to warranted awareness (prama¯).
Buddhist logicians identify two sources of knowledge: perception (pratyaks. a) and inference (anuma¯na). Perception is said to be an immediate contact with particulars without any mediation of conception (which is considered to involve universals). There are two standard examples used to illustrate inference. In one standard example, when we are aware that there is smoke on a mountain, we may infer that there is ﬁre on that mountain. When we become aware of the presence of ﬁre on the mountain in this way, that awareness is said to be warranted by inference and, thus, it counts as knowledge. In another example, we may become aware of the presence of a tree by inferring from our awareness that there is rosewood (s´im. s´apa¯). Such awareness of the presence of a tree that is brought about by inferential cognition is said to be warranted and, thus, it is ascribed the status of knowledge.
Digna¯ga explains that the purpose of his texts — the foundational texts for the Buddhist epistemological and logical tradition — is to refute his opponents’ views on the instruments of knowledge as well as to establish his own view as correct.5 Once inference is shown to be an instrument of knowledge, it is also shown to serve as an instrument for producing a cognition that can be ascribed the status of knowledge. Hence, inference can be thought to result not only in awareness that inference is an instrument of knowledge, but also in awareness of the truths of Buddhist thought.
Buddhist logicians recognise two contexts in which inference can be used as an instrument of knowledge. On the one hand, it can be used as an instrument for becoming aware of soteriological truths, such as the four noble truths, by them-selves. This is called ‘inference for oneself’ (sva¯rtha¯numa¯na). On the other hand, inference can be used as an instrument in dialectical engagements with opponents.
In this context, inference serves as a tool for showing that the opponents’ views are mistaken and to demonstrate that Buddhists’ own views are correct in dialectic practice. This is called ‘inference for others’ (para¯rtha¯numa¯na). Thus, inference has a function directed at oneself and a function directed at others.
If we understand inference as a source of knowledge, it must be assumed that a certain relationship holds between inference and knowledge. In particular, it must be assumed that inference has a direct impact on one’s knowledge state. As is im-plicit in the standard examples, inference subsumes three elements which might be thought to be distinct. One element of inference in the standard examples is the awareness of smoke or rosewood. This is the cognitive state prior to the involve-ment of the inferential cognitive process. The second element is the inferential cognitive process which moves cognition from awareness of smoke or rosewood to awareness of ﬁre or tree. The third element is the resulting cognitive state, that is, awareness of ﬁre or tree. These elements of Buddhist logic may strike some
as problematic given that ‘inference’ (or logic in general) is often understood as not having any bearing on knowledge. Some may even consider them to show that there is no logic in the Buddhist tradition (as we will see below). I will con-sider this issue before going on to explicate how Buddhist logicians understand inference in the context of epistemology.
According to the ‘Western’ conception of logic prevalent since the twentieth century, logical study is the formal study of arguments. An argument consists of premises and a conclusion (or conclusions). A study of logic is considered to be a study of the (syntactic) form of the relationship between the premises and a conclusion. If an argument has the form: A and A � B (if A then B), therefore B (modus ponens), for example, then the argument which instantiates this form is said to be valid. The form that an argument instantiates may also capture a sub-sentential (sub-propositional) structure. For example, an argument may have the following form: 9x8yRxy (Something Rs everything), therefore 9xRxx (Some-thing Rs itself). Any argument which has this (sub-sentential) form is also said to be valid.
Western logicians disagree about which forms of argument should count as valid. (Strict) Aristotelian logicians regard the above sub-sentential structure (ﬁrst-order structure) invalid as it is not one of the syllogisms that Aristotle identi-ﬁed. Intuitionistic logicians reject as invalid the form: ::A (it is not the case that A is not the case), therefore A, as they interpret negation in terms of the failure to ﬁnd a proof, and, thus, a failure to ﬁnd a negative proof doesn’t tantamount to a success of ﬁnding a positive proof. Paraconsistent logicians — whose logics have recently been applied to the study of reasoning involving catus. ko.ti — may regard modus ponens invalid. If A were both true and false, then A and A � B would be true (since A would also be false) while B might be only false.7 Which forms of argument should count as valid is evidently a contentious mat-ter. Nonetheless, there is one thing that contemporary (western) logicians share in common — namely, the idea that logic is concerned with the forms of argu-ments. An argument is said to be valid if it instantiates a valid argument form.
It is important to notice that, for contemporary logicians, the speciﬁc contents of the premises and the conclusion are irrelevant to logical study. The main focus of logical study is with the mathematical structures that satisfy valid argument forms rather than a concern with what the argument is about.
If we understand the nature of logic to be formal in this way, it is di�cult to see what bearing logic has on cognitions that result in knowledge. Consider the following argument: there is smoke on a mountain, and if there is smoke on a mountain then there is ﬁre on that mountain, therefore there is ﬁre on that mountain. This argument has the form of modus ponens and, thus, it is formally valid. Now, assume that we know that there is smoke on the mountain and that where there is smoke there is ﬁre, but we don’t know that there is ﬁre on the mountain. It seems that we ought to come to reason so that we know that there is ﬁre on the mountain.
However, what is the role of the validity of modus ponens in this reasoning? In order to think of modus ponens itself as having an impact on our cognitive activity, it wouldn’t be su�cient for modus ponens to be, in fact, valid: we would have to be aware of the validity of modus ponens. We would also have to be aware that if an argument has a valid form, then we would have to infer the conclusion of the argument. But, in order to have this awareness, we would also have to be aware that if we are aware that ‘if an argument has a valid form, then we have to infer the conclusion of the argument’, then we would have to infer the conclusion.
But, then, we would also have to be aware that ... ad inﬁnitum. It seems that the form of the argument is not su�cient for thinking that there is ﬁre on the mountain.9 The di�culty of accounting for the relationship between valid forms of argu-ments and knowledge can also be illustrated by examining what kind of cognitions are, in fact, involved in order to acquire new knowledge. We most likely infer that there is ﬁre on a mountain if we are aware that there is smoke on that mountain and that where there is smoke there is ﬁre. But, that is because awareness that there is ﬁre on a mountain is consistent or coherent with what we already know, and not because cognitive activity involved in inferring that there is ﬁre on a mountain has the form of modus ponens.
When we acquire new knowledge by means of inferential cognitions, it is not the form of an argument (or reasoning) that forces us to infer the conclusion but the consistency (or coherency) of what we are aware of that moves us to cognise in a certain way. Hence, it seems that knowledge of the validity of an argument form is not even a necessary condition for knowing that there is ﬁre on a mountain based on what we already know.10 Faced with these di�culties in accounting for the signiﬁcance of epistemol-ogy in formal logic, we have (at least) two options in our attempt to understand
the role that inference (anuma¯na) is said to play in Buddhist epistemology. One is to suggest that the study of anuma¯na is not a study of logic at all (and, thus, it is misleading to translate it as ‘inference’). After all, anuma¯n. a is an instru-ment for acquiring knowledge and, as such, it is an epistemological, not a formal, apparatus. Thus, according to this suggestion, there is no logic in Buddhist episte-mology; there is just epistemology.11 It would follow that, strictly speaking, there is no Buddhist philosophy of logic.
This is not the only way in which we can respond to the above di�culties, however. Alternatively, we can re-conceive the nature of logic in the context of epistemology and demarcate the logical part of epistemology which can be recog-nised as logic.12 To think of an inference simply as a formal matter is to think of it as having logical signiﬁcance independent of the cognitive act which can be characterised as inferential. If we think that logic is only about argument forms, it is di�cult to understand the impact of inference on cognitive activities. However, we can understand inference as having logical signiﬁcance within certain cogni-tive acts of inferring (or cognition which acts inferentially). Thus, according to this suggestion, we can focus on the nature of the role that inferential cognition plays. This is not to say that the form in which such a cognition takes place cannot be ascertained. It does suggest, however, that the form of a cognition is not what is at issue for Buddhist logicians.
Which of these two suggestions is most faithful to the Buddhist tradition is a controversial matter that I shall not attempt to settle in this chapter. If we are to elucidate Buddhist philosophy of logic, however, we must assume that we can recognise logic as part of the Buddhist study of anuma¯na (inference). Based on this (defeasible) assumption, I present what I take to be the conception of logic assumed by Buddhist logicians. Since logic is conceived in the context of epis-temology, the Buddhist conception of logic must be unique (at least in compari-son with twentieth- and early twenty-ﬁrst-century Western orthodoxy). If so, this chapter does not merely present the way in which Buddhist logicians understand the nature of the object of their study, it also makes an important contribution to the philosophy of logic in general in that it develops an alternative to the concep-tion of logic as purely formal in nature.
The Elements of Inference
Buddhist logicians hold that, by inference, one seeks to establish for oneself or for others the presence of what is to be proven (sa¯dhya) in a particular locus (paks. a) on the basis of the presence of an inferential reason (hetu) at that locus. For example, to use a standard example, when one sees smoke (inferential reason) on a particular mountain (locus), one may infer that there is ﬁre (what is to be proven) on that mountain.
For an inference to count as ‘valid’ or an instrument of knowledge, the in-ferential reason must satisfy the triple conditions (trairu¯pya). There are mainly two interpretations of these triple conditions: ontological and epistemological.
According to the ontological interpretation (which seems the more common inter-pretation), they can be presented as follows. (1) What is identiﬁed as the inferen-tial reason must be present in the locus in question. This means, in our example, that smoke must be present on the mountain in question. This implies that a cog-niser is, in fact, in a situation where there is smoke on the mountain. The second and third conditions are called pervasion (vya¯pti). (2) The inferential reason must be present in at least one similar case (sapaks. a) — a locus where what is to be proven is present.14 For example, smoke must be present in a kitchen with a wood-burning stove. (3) The inferential reason must not be present in any dissim-ilar case (vipaks. a) — a locus where what is to be proven is absent. For example, smoke must not be present in a misty lake. The reason given for a thesis to be proven by inference must satisfy these triple conditions to count as an instrument of warranted awareness (i.e., knowledge).
According to (a strong form of) the epistemic interpretation, the triple con-ditions can be presented as follows: (1) the inferential reason must be known to occur in the locus, (2) the inferential reason must be known to occur in a simi-lar case, and (3) the inferential reason must be known not to occur in dissimilar cases.15 The di�culty of reconciling the ontological and epistemological interpreta-tions is due to the epistemic context in which inference must be understood. In order to properly understand inference in epistemic contexts, we cannot think of these two interpretations mutually exclusive. According to the ontological in-terpretation, it is a fact that the triple conditions are satisﬁed.
This fact can be appealed to as evidence for justifying one’s awareness of the presence of ﬁre on a mountain, for example. That is, if it can be demonstrated that one’s awareness of ﬁre can be characterised by the triple conditions, then that awareness, not its form but its content, can be given the status of knowledge by appealing to the fact about the satisfaction of triple conditions. This means that the triple conditions function as the standard of evaluation for what counts as knowledge.
If one ac-cepts this ontological interpretation, however, it is di�cult to understand how the triple conditions have bearing on the acquisition of knowledge. Just as the validity of an argument form is not su�cient for coming to have knowledge, the fact that there is a justiﬁcation doesn’t seem to be enough for one to acquire knowledge.16 If we are to think that the triple conditions serve as a means of knowledge, they must contain the epistemic fact about one’s having knowledge. Hence, if we are to understand the role that inference plays in epistemological contexts, we need to accommodate both intepretations.
Our discussion in the previous section suggests that inference is both ontological and epistemoplogical. As an instrument of warranted awareness, inference must warrants or justiﬁes one’s awareness but must also be instrumental in moving one’s cognitive state to another cognitive state.
onsider, again, our example of smoke and ﬁre on a mountain. Suppose that one is in a situation where there is smoke on a particular mountain. In order for awareness of ﬁre on that mountain to be warranted, smoke must co-occur or must be co-located with ﬁre on the mountain in the same way that smoke co-occurs with ﬁre in a kitchen. However, smoke on a mountain must never co-occur with ﬁre in a misty lake despite the fact that it may look as though there is smoke on the lake and, thus, there may be a mistaken ‘appearance’ of ﬁre in the lake. Supposedly, these are the facts about the elements of inference. It may be di�cult to establish that these facts, in fact, obtain. Nonetheless, if they are presented as facts, one can appeal to them in order to justify the inferential awareness of the presence of ﬁre on the mountain based on the awareness of the presence of smoke on that mountain.
For pervasion between inferential reason and what is to be proven to have any cognitive grip, however, it has to be expressed to us in an intelligible manner. For example, I would not be cognitively moved at all by the mere fact that there is ﬁre where there is smoke, unless that fact places some demand on my cognition.
Nor would my cognition be stimulated to produce awareness of the presence of ﬁre if I am simply perceiving smoke on a mountain. An immediate, perceptual awareness of smoke is not, of itself, enough for awareness of the presence of ﬁre to warrantably occur. In order for cognition to be warranted and a cognitive process required such that certain awareness is brought about, we must connect the concept smoke with ﬁre in a cognitively robust relation. By making a conceptual commitment to invoke ﬁre together with smoke, we can act appropriately in the presence of ﬁre when we become aware of smoke. The resulting awareness is dependent upon the warrantable relation of the elements in the inference becoming cognitively signiﬁcant.18 The concepts invoked in making a conceptual commitment are not fully formed independent of their use in cognition. Instead, the contents of concepts are shaped by the commitment one undertakes through inferential cognition. It is by making a conceptual commitment that we acquire knowledge. In turn, the triple con-ditions which hold between the elements of inference become incorporated into cognition.
As the content of a concept becomes more determinate, it is assumed that one may come to recognise more robust relations between smoke and ﬁre. The more one learns about smoke (its chemical constitution, etc.), the better one ascertains its relation to ﬁre. Of course, there may be a situation where we may have to revise our commitment about smoke and ﬁre. If one thought that smoke is just something caused by cooking, for example, then one might have to revise one’s epistemic commitment so that one would now be aware that smoke is something which is caused by ﬁre. So an inferential cognition may defeat the inferential cognition that forms the content of smoke in its premature form (when one wasn’t aware that smoke is caused by ﬁre and not by cooking as such).19 In this way, as the content of concepts becomes more determinate through inferential cognitions, one attains more certainty as well as truth of the matter and, thus, acquires warranted awareness.
One can provide mathematical structures that satisfy these triple conditions as a study of Buddhist ‘logic’.20 They may express the forms that inferential cognition may take. Nonetheless, for Buddhist logicians, inference is considered to be im-portant insofar as it shapes the contents of our awareness. That is, the importance of inference lies not in ascertaining forms of knowledge but in their contribution to the content of knowledge. It is in this way that we can see an understanding of the nature of inferential cognition as a crucial aspect of Buddhist epistemol-ogy as well as the conception of inference (and logic more generally) that can be attributed to Buddhist logicians.
If we thought that logic was essentially formal, the notions of inference and infer-ential cognitions must be assumed to come apart in a certain way. An inference might be said to underwrite the validity of an inferential cognition. Yet it would not impel one to undergo an inferential cognitive process. According to the for-mal conception of logic, an inference expresses the fact about the validity of an inferential cognition, but it doesn’t express the norm by means of which an infer-ential cognition ought to take place. Thus, under the formal conception of logic, an inferential cognition must be understood as independent of inferences.
Buddhist logicians, by contrast, don’t think of inferential cognitions as in-dependent of inferences. An inference expresses an epistemic commitment that forms the content of a warranted awareness. This means that Buddhist logicians don’t think of inference as underwriting an inferential cognition. Rather, an infer-ence is considered to express an epistemic commitment, particularly a conceptual commitment, that is undertaken in an inferential cognitive process. For Buddhist logicians, the signiﬁcance of inference lies within the context of inferential cog-nitions.
Thus, the Buddhist study of inference (anuma¯na) is not a formal study of argu-ments. Buddhist logicians are not concerned with abstracting forms of arguments and studying the (mathematical) properties of these forms. While one might take this as an evidence for the thought that there is no Buddhist logic, one might alter-natively attempt to understand the role of inference according to a conception of logic that does not abstract from the contents of knowledge.21 As I have attempted to show, we can recognise the logical signiﬁcance of inference as understood by Buddhist logicians despite the fact that its logical signiﬁcance lies within the con-text of (contentful) cognition.