The Character of Logic in India (review)

Few subjects are as bewilderingly complex as the topic of Indian theories of knowledge and valid inference. They are a thicket of scholastic intrigue, replete with enough arcane distinctions to drive the most hardened logician to exhausted despair. Consequently, it is a genuine accomplishment, if not a universally recognized one, to succeed in actually sorting out these discussions and bring the light of intelligibility and relevance to the subject. The late Bimal Krishna Matilal left us a fitting legacy in this small, unfinished, and posthumous masterpiece of clarity, The Character of Logic in India. Surveying the development of Indian accounts of valid inference, Matilal attempts essentially two things. One is, of course, to identify and explain the distinctive aspects of such accounts. This by itself is a formidable challenge due to the complexity of the Indian philosophical scene with its various traditions of scholarly discourse, whether they be Buddhist, Jaina, or the "Six Schools." But Matilal also attempts to say something both comparative and general concerning the nature of logic itself, and this, to my mind, is even more of a challenge. For one thing, logic by itself is a difficult subject, where divergences in the basic operating vocabulary make a profound difference. Additionally, however, while it was popular as a topic for philosophical analysis some years ago, logic has descended in importance and now appeals to an ever narrower range of philosophers. Thus, a project such as this involves the double challenge of obscurity. My private hope is that not only will works such as this one give comparative philosophy a helpful boost, but that the study of logic, and inference in particular, will be encouraged as well.

Matilal begins his book with a kind of overview of Indian accounts of valid inference as well as a general discussion of its differences with the Western tradition of philosophical logic. Most of the major themes of the book emerge at this point. First of all, according to Matilal, the study of logic is the study of valid inference, that is, what allows us to draw a particular conclusion from a premise or premises. From a Western perspective this becomes a question of how certain propositions are related; from an "Indian" perspective it is more a question of how one sort of knowledge can give rise to another. That is, "Indian logic" answers the question "When is inference a valid form of knowledge?" and so combines both validity and soundness in the same approach. Matilal characterizes this as "psychologizing" or "epistemologizing" logic but defends it on the grounds that it is still a concern with the objective content of the elements of logical inference, not with the psychological connection of mental states. This is psychologizing in the "bad" or, I suppose, nonlogical sense. The fact that Indian logic combined soundness and validity also points to another fact: Indian accounts of inference cover not just deductive connections but inductive ones as well. That is, there is no systematic difference between what is to count as a specifically inductive as opposed to a deductive inference. Again, Matilal tells us that this is due to the fact that Indian logicians, whether Buddhist, Nyāya, or Jaina, were principally interested in "knowledge-warranting" relationships between various sorts of evidence.

The epistemological thrust of Indian logic, as Matilal tells it, also explains some of its more unique and often puzzling features. For example, consider the emphasis on always having both a specific example of some general claim made in the space of an argument and a specific "negative" example. In fact, the Buddhist Dinnāga, in one of the earliest accounts of logical inference, insisted on the presence of both example (sapaksha) and counterexample (or better, counter-positive example) (vipaksha) in a valid inference. The reason is that if one provides an example of a universal claim such as "all crows are black," then the problem of vacuously true propositions is avoided and the...