UBC Theses and Dissertations
The Navya-Nyāya theory of inference Mullatti, Laxman Channappa
This dissertation is an attempt to represent the Navya-nyaya (Nn) theory of inference in terms of a contemporary conceptual framework. There exist similar studies by able modern scholars, but these are either piecemeal, or, when comprehensive, rely mainly on traditional (western) logic. The present attempt is both relatively more comprehensive, and employs notions from contemporary logic. Since the Nn theory of inference is couched in terms of cognitions rather than sentences, an effort is made to clarify the notion of cognition. It is argued that a cognition is an abstract and ontologically independent entity much in the fashion of the Fregean proposition, and that primarily it, as distinct from its linguistic vehicle, is the bearer of truth-values. The semantic model set up by the Navya-naiyayikas (NNs) for a cognition is considered, and its inadequacy brought out with reference not only to universal cognitions, but also to certain modes of expression generally accepted as synonymous. These modes of expression also serve, in part, to bring out the inadequacy of the linguistic criteria that the NNs implicitly use for individuating cognitions. A further defect of the Nn semantic model is that it requires that n-adic predicates be reducible to monadic ones, a requirement that can hardly be fulfilled. Partly in the course of their search for a principle of individuating cognitions, the NNs bring in three notions, namely, expectancy, competency, and proximity. Of these, expectancy is viewed as a syntactic property, and compared to a certain communication theoretic model for linguistic structure employing a finite state Markov process. Competency is sought to be understood with reference to Ryle's theory of category mistake, and, in the process, the difficulties of the thesis that unexampled (non-referring) terms lack competency are pointed out. Proximity is regarded primarily as an attempt at introducing some kind of word-order, especially into the shorter sentences of Sanskrit (which, strictly speaking, i.e., from a grammatical standpoint, have no rigid order). Next, the Nn definition of inference—that inference is the instrumental cause of the conclusion—is examined, and the usual way of understanding it in a psychological sense is shown to be unsatisfactory. Inference is viewed as a set of cognitions such that the conclusion is a logical consequence of the set of the remaining cognitions. A case is made for understanding the term 'cause' in the above definition in a logical sense to mean an inference-form. It is shown that, on such an understanding, inference for oneself and inference for others collapse into one, and that both have exactly the same number of cognitions, namely, four, as elements. The resulting discrepancy between the number of sentences and the number of cognitions in inference is accounted for by means of non-logical considerations. The nature of each of the four elements is explained with particular emphasis on the second element, pervasion. The representations by Staal and Berg of the Nn definition of pervasion are considered and rejected, and a much simpler representation proposed. It is pointed out that the Nn distinction between positive and contrapositive pervasions involves a recognition, with certain reservations, of the (complete) law of contraposition. The third element of inference, 'consideration', is shown to be superfluous, and Schayer's interpretation of it erroneous. It is argued that the elliptical expressions used to express the third and the fourth elements are intelligible only in relation to the early stages of Nyaya when the concept of pervasion as a member of inference had not yet evolved, and that they cease to be intelligible in relation to the later stages when pervasion is included as an element. This fact is used to point out the untenability of the suggestion that the Nn syllogism is really the Aristotelian argument from example. The paradoxes of relativising pervasion to the inference containing it are brought out, and the theory of confirmation implicit in the Nn account of pervasion is fully reconstructed with special reference to Hempel and Goodman. The Nn account of another (non-syllogistic) form of inference is considered, and Staal's representation of it shown to be wrong. Finally, it is observed that the NNs do not call in question the formal validity of an inference, and that, for them, a fallacy is necessarily 'material'. It is also maintained that the customary view that the Nn notion of accident is a means of converting an unsound inference into a sound one is mistaken, and that the true function of an accident is to show up the falsity of a pervasion.
Item Citations and Data