ONTOLOGICAL CATEGORIES IN EARLY INDIAN PHILOSOPHY
1. those that belong to the Emperor,
2. embalmed ones,
3. those that are trained,
4. suckling pigs,
6. fabulous ones,
7. stray dogs,
8. those included in the present classification,
9. those that tremble as if they were mad,
10. innumerable ones,
11. those drawn with a very fine camelhair brush,
13. those that have just broken a flower vase,
14. those that from a long way off look like flies.
It is probably not necessary to remind you that this classification does not occur in any real Chinese Encyclopaedia, that it is the product of the imagination of a famous author, Jorge Luis Borges (Argentina, 1899-1986), who used it in his essay "The Analytical Language of John Wilkins". This hardly matters for our present purposes, for it reminds us that the usefulness of categories depends of what we expect from them. There are less exotic categorizations that are useful in one context but less so in another. In certain situations it makes perfect sense to distinguish between animals that live in the sea and those that live on land. Whales and dolphins will in this way be categorized as animals that live in the sea, like fish. There is nothing wrong with this categorization, except that biologists will prefer another one, in which whales and dolphins are mammals and must therefore be categorized with many terrestrial animals. Biologists prefer this second categorization, because they believe it provides information that reaches below the surface: the biological categorization tells us something about the place of whales and dolphins in the animal kingdom, about their ancestry and relatedness to other animals.1
These reflections suggest that some categorizations seem more “real” to us than others. We think that the categorization of animals in the Chinese Encyclopaedia is rather useless because it does not tell us anything more about these animals and their relationships to each  other than the mere fact of being listed here. The example from biology further supports the idea that some categorizations are indeed more “real” than others, that they somehow reveal an aspect of the world that might otherwise escape us. The categorization of whales and dolphins as mammals is, as a matter of fact, the result of biological research. Before this research had been done, they would no doubt have been wrongly categorized. Thanks to the knowledge acquired in biology, they can henceforth be categorized more satisfactorily, and presumably more “correctly”.
If I were a philosopher speaking about the philosophy of science, I might conclude from all this that categorizations are risky, that they may have to be modified in the light of new knowledge, that rigid categorizations should be avoided before all the data are in, etc. However, I do not speak about the philosophy of science, nor am I a philosopher. I am an indologist trying to understand how and why categories were used in early Indian philosophy.
Since I wish to speak about the conscious and explicit use of categories, the only forms of Indian philosophy that have to be considered are its systematic manifestations. Indian systematic philosophy, I have argued elsewhere, began among Buddhists in the Northwest of the subcontinent, and subsequently made its appearance elsewhere, also among Brahmins and Jainas. The first attempt at creating a coherent system of thought is associated with the Sarvastivadins, and it clearly had its ontological categories.
However, the Sarvastivadins did not have to create their ontological categories. They were Buddhists, and were therefore more or less bound to the categories that were part of this tradition. The list of five skandhas, in particular, was there, had always been there, and had been accepted by the Buddha, so this list had to be valid and validated. (Just to remind you, the five skandhas are known by the names rupa, vedana, samjna, samskara and vijnana respectively.) The ontological attempts of the Sarvastivadins had to find a place for them, and it turned out that they could be used as categories for dharmas.
I do not need to remind you that the ontological attempts of the Sarvastivadins were elaborations of their decision to take the lists of dharmas that had been collected by Buddhists for some time as lists of elements of existence. Human beings, and everything else, are in the end no more than accumulations of dharmas. Strictly speaking, only dharmas exist; accumulations of dharmas have no separate existence.
The five skandhas had always been presented as the five constituents of human beings. It is easy to see that the joint observation that human beings are accumulations of dharmas and that human beings consist of the five skandhas led to the conclusion that each dharma belongs to one skandha, and that the five skandhas are therefore a way to categorize the dharmas. This, as we know, is what these Buddhists did: a traditional list of dharmas was categorized with the help of what was thought of as a traditional list of categories.
None of this is very remarkable from the point of view of our interest in categorization. What is remarkable, is that these same Buddhists became dissatisfied with their traditional categories, and created another list of five. To use the terminology of Frauwallner,2 who drew attention to this development, a newly invented Pancavastuka replaced the traditional Pancaskandhaka. The five vastus of the Pancavastuka are: rupa, citta, caitasika dharma, cittaviprayukta samskara, and asamskrta.
The advantage for the Sarvastivadins of the Pancavastuka over the Pancaskandhaka is undeniable. The new categorization left place for a number of dharmas which were crucial in the ontology elaborated by these Buddhists. Two of the five categories of the Pancavastuka — that of the cittaviprayukta samskaras (non-mental dispositions) and that of the asamskrta dharmas — had nothing corresponding to them in the Pancaskandhaka, yet contained a number of dharmas that were essential to Sarvastivada ontology. From the point of view of  the Sarvastivadins, the Pancavastuka was more “real” than the Pancaskandhaka, closer to reality as they conceived of it.
Unfortunately, it was also a new invention, one difficult to justify on the basis of the words traditionally attributed to the Buddha. Frauwallner regards “the creation of the Pancavastuka as the most important step on the way from Buddhist dogmatics to a philosophical system” (p. 147). He may be right, but the fate of the Pancavastuka illustrates that a Buddhist philosophical system is never just a philosophical system; it is also a religious tradition. The Buddhist tradition did not feel comfortable with this new invention, and preferred the Buddha's words to the inventions of his followers. This explains that, even within Sarvastivada and in spite of a promising beginning, the Pancavastuka came to play a secondary role in subsequent treatises, including Vasubandhu's Abhidharmakosa and Bhasya. Obviously the Buddhist philosophers were all in favour of categories, on condition that the Buddha had pronounced these categories.
There are good reasons to believe that the Brahmanical ontology that goes by the name Vaisesika was created in response to Sarvastivada ontology. Both can to a surprising extent be presented as answers to the same questions. The questions were the same, but the answer were different. Whatever the details of the development that led to the formulation of Vaisesika ontology, its makers had a much freer hand than the Buddhists. They did not believe that this ontology was already present in their textual tradition — primarily the Veda —, nor did they think that a complete enumeration of all that exists was to be found there. They were free to create an ontology ex nihilo, so to say.
However, they had to respect one important restriction. Brahmanism is a tradition of the word. The word — i.e., the Sanskrit language — is eternal, and finds its purest expression in the Veda. The Sanskrit language is not just any language; in an important sense it is the only language, of which all other languages are no more than corruptions. This one and eternal language is also close to reality. Indeed, the idea that the world is created at the beginning of each cosmic cycle in accordance with the words of the Veda is a recurring mythological theme.
The Vaisesika philosophers took the close connection between words and things for granted. The argument that this or that “thing” must exist because there is a word for it recurs several times in their writings. As a first approximation, one can therefore say that a list of all Sanskrit words is a list of all existing things. This approximation has to be refined, to be sure, but we can easily see that these Brahmanical thinkers did not need a list of dharmas which had provided their Buddhist confreres with an enumeration of all that exists. Their sacred and eternal language provided them with much of what they needed.
We saw that the Buddhists categorized their dharmas with the help of other traditional lists, which they looked upon as lists of categories. The list of five skandhas was considered particularly useful for this purpose, but we also saw that the Sarvastivadins were confronted with its shortcomings and tried to improve upon it. The Vaisesika thinkers had no traditional lists to choose from and could only draw inspiration from the Sanskrit language. Does the Sanskrit language provide ontological categories?
It does, at least in the understanding of the early and highly respected grammarian Patanjali's, author of the Great Commentary (Mahabhasya) on Panini's grammar (soon after 150 BCE). Patanjali's distinguishes three kinds of words, which he calls jatisabda, gunasabda and kriyasabda respectively.3 These correspond to our nouns, adjectives and verbs, and designate substances (dravya),4 qualities (guna) and actions (kriya) respectively. The Vaisesikas took this to heart, and divided all, or almost all, existing things into these three categories: substances, qualities and actions.5 For reasons internal to their system they added a few more categories, so as to arrive at six, seven or ten, but their fundamental categorization consisted of these three; the other  ones were mere add-ons to deal with theoretical questions that came up.
Why did they choose these categories, not any others? I do not think that
Vaisesika texts ever raise this issue in its generality, and it is easy to understand why. A difference of opinion about some of the added categories might be possible, and did indeed occur. The core of their system of categorization, on the other hand, was not open to debate because it was a given. It was obvious to a Vaisesika that in a situation described by the words “the blue bird ate the yellow butterfly”, two substances are referred to (the bird and the butterfly), two qualities (blue and yellow, which reside respectively in the bird and the butterfly), and one action (eating). This, he might have argued, was not a categorization imposed upon reality, but a given. In other words, the Vaisesika categories — or at any rate the ones they all agreed on — constitute a natural categorization. It tells us something about reality (just as the categorization of whales as mammals tells us something about reality), and is not an option that could fruitfully be replaced by another categorization.
If, armed with this information about categorization in early Indian ontologies, we turn to the theme of this conference (“The Idea of a Category in Indian Philosophy”), we have to ask ourselves what ideas the Sarvastivadins and the Vasesikas had of the categories they imposed upon the world. Of the categorizations considered so far, the Pancavastuka might be expected to be the most interesting, for it was admittedly a human creation invented to fit the ontological data accepted by its creators. The other categorizations considered, the Pancaskandhaka and the fundamental categorization of Vaisesika, are less promising, for they were not believed to have had any creators in the ordinary sense: the Pancaskandhaka owed its acceptance to the fact that the Buddha was believed to have pronounced it, and the fundamental categories of Vaisesika were looked upon as a given that needed no creator. Without a creator, they needed no justification either.
What about the Pancavastuka?
We have seen that the Pancavastuka did not come to occupy the place which its qualities should have earned for it. We have also seen that the reason why it had to content itself with a secondary position in Buddhist ontological discussions was the very fact that ordinary humans had invented it. It is one of the tragedies of the history of Indian philosophy that these “ordinary humans”, who had created a new categorization, were not given the respect by their successors which in our opinion they deserved. No discussion arose about the need to invent a new categorization if the traditional one is not satisfactory. No questions were asked about the place and justification of categories in general. This is a pity, for it makes it all but impossible to answer the question of our conference as far as ontology in early Indian philosophy is concerned.
References: Bronkhorst, Johannes 2004. “La grammaire et les debuts de la philosophie indienne.” Asiatische Studien /Etudes Asiatiques 58(4), 791-865.
Dawkins, Richard 2009. The Greatest Show on Earth. The evidence for evolution. London etc.: Bantam Press.
Frauwallner, Erich 1963. “Pancaskandhaka und Pancavastuka.” Wiener Studien fur die Kunde Sud- und Ostasiens 7, 20-36. English translation: Frauwallner 1995: 135-147.
Frauwallner, Erich 1995. Studies in Abhidharma Literature and the Origins of Buddhist Philosophical Systems. Albany: State University of New York Press.
Jones, Steve 1999. Almost Like a Whale. The Origin of Species updated.
Doubleday. Black Swan edition, 2001.
Maha-bh Patanjali's, (Vyakarana-)Mahabhasya, ed. F. Kielhorn,
1 Steve Jones dedicates a chapter of his book Almost Like a Whale to the importance and significance of classification in biology (1999: 364 ff.). Note that ‘fish' are not a natural group: “To an evolutionist ... a ‘natural' group of animals is a group all of whose members are closer cousins to each other than they are to all non-members of the group. ‘Birds' . are a natural group, since they share a most recent common ancestor that is not shared by any non-bird. By the same definition, ‘fish' and ‘reptiles' are not natural groups. The most recent common ancestor of all ‘fish' is shared by many non-fish too.” (Dawkins 2009: 162).
2 Frauwallner 1963; English translation: Frauwallner 1995: 135-147.
3 Maha-bh I p. 20 l. 8-9 (under Sivasutra 2) : traylca sabdanam pravrttih /jatisabda gunasabdah kriyasabda iti /.
4 Maha-bh I p. 230 l. 17 (under P. 1.2.58 vt. 7) : jatisabdena hi dravyam apy abhidhlyate jatir api.
5 Bronkhorst 2004.