09 May 2014

Where and Why Did the Sarvāstivādins Go Wrong?

image: wisegeek
It's widely thought that both the Perfection of Wisdom texts and the writings of the Madhyamaka School attack the Realist position taken by the Sarvāstivāda School. The difficulty we have at 2000 years remove is understanding how any Buddhist could adopt a Realist position in the first place. Surely the Middle Way would have ruled this out?

However, as I showed in my essay about Action at a Temporal Distance, early Buddhists inherited a major problem: pratītyasamutpāda and karma as outlined in the early Buddhist texts are inconsistent with each other. I showed that Buddhist schools modified the doctrine of pratītyasamutpāda to preserve the doctrine of karma more or less as it was.  In the case of Sarvāstivāda, this solution required that dharmas be able to function as conditions -- tantamount to being real -- in all three times: present, future and past. But this was not the only influence pushing Sarvāstivādins towards Realism.

As with other early Buddhist schools, the Sarvāstivāda focus moved onto the Abhidharma project, with Abhidharma texts quickly attaining canonical status. Each of the surviving Abhidharma texts is distinct in it's content, if not in its overall project and methods. Thus Abhidharma is a product of sectarian Buddhist schools which often see each other as rivals. 

The key task of the Abhidharma is to identify dharmas, to catalogue and describe them and to explore the dynamic relationships between them. Colette Cox, one of the leading writers on the Sarvāstivāda, calls these two functions: evaluative and descriptive (2004). Ābhidharmikas evaluated dharmas for their contribution to liberation and used the descriptive analysis of dharmas to deconstruct perceived structures and realities (particularly the self). Cox notes that this analysis became increasing fine grained and abstract. This in turn created the conditions for treating dharmas as real entities (dravya). This essay will explore this second Realist influence on early Buddhism, again focussing on the Sarvastivāda as representative of Indian Buddhism.


Collette Cox (2004) sets out the process by which Sarvāstivādins grew into the idea of dharmas as real entities (especially 558-565). Since one of the main functions of the Abhidharma was to create a taxonomy it had to take the approach that all taxonomy projects must: it had to create categories, and criteria by which any dharma might fit into any category. The Sarvāstivādins concentrated on a method called "inclusion" (saṃgraha) in which for each dharma, they outlined what categories it fit into. Each dharma might fit into multiple categories, but it either fit or did not. 

All taxonomic projects have to proceed in a particular way in order to create meaningful and useful categories. And one of the main features of such projects is well defined categories, "...invariable criteria are demanded as the basis of unambiguous classification" (Cox 59). Human beings think about the world using categories. Contemporary understanding of these categories shows them to be based on resemblance to a prototype that sits somewhere in the middle of the taxonomic hierarchy (Lakoff 1990). The edges of such categories are fuzzy and membership is by degrees.

Categories are an efficient way of dealing with large amounts of information and also to assessing the potential of a new entity or event by seeing it's similarity to familiar entities or events. And Lakoff argues that the categories we use are in part defined by how we interact with the members of the category - either physically or metaphorically (where the source domain for the metaphor is itself a physical interaction).

The categories used by Ābhidharmikas, by contrast, seem to have been hard edged and made up of simple, artificial binaries and trinaries. For example a dharma was either samskṛta or asamskṛta; either kuśala or akuśala; etc. A Buddhist needed to know kuśala from akuśala dharmas for the purposes of pursuing liberation (cf Cox's two functions above). Kuśala dharmas are to be cultivated and akuśala dharmas to be abandoned. Matrices (mātṛka) of binary and trinary categories are thought to have made up the earliest Abhidharma "texts". Lists were memorised by the mātṛkadhāra (the counter part of the sūtradhāra and vinayadhāra) and the whole structure of Abhidharma categories could be (re)constructed from such lists. The descriptive enterprise of deconstructing apparent entities, particularly the self, into constituent dharmas gave rise to an encyclopaedic attempt to list all possible dharmas and the possible relations between them. Categories multiplied until we get the lists of 85 dharmas in the Theravāda and 75 in the Sarvāstivāda Abhidharmas respectively.

Each school of early sectarian Buddhism took a slightly different approach to constructing categories, placing dharmas in them, and elaborating their relations. This suggests that the impetus itself is pre-sectarian and indeed we can see examples of Abhidharma style thinking in the early sūtras, as well as as a common core in early Abhidharma texts. However, the manifestations which survive as canonical texts and commentaries are a product of sectarian Buddhism.

A dharma fits into a particular category because it has a particular function or nature which is referred to as svabhāva. This term originally had no ontological implications. Initially svabhāva is simply what enables us to distinguish one dharma from another (Cox 561-2).


The early Buddhist worldview centred around the idea that experience is a flux of conditional processes, arising and passing away as our minds and sensory apparatus are impacted by objects, mental and physical. In this view a dharma was primarily the object of the mind sense (manas), though of course the word dharma is confusingly used in at least six main senses (see Dharma: Buddhist Terminology). But in relation to Buddhist thinking about experience, a dharma is a mental object. In this sense, perhaps, dharma comes closest to it etymological meaning of "support, foundation" - a dharma is the "objective" support process for a mental event to arise (i.e. for an experience to be conscious) when it interacts with the "subjective" perceiving processes. 

The first step on the road to seeing dharmas as real seems to have been the fixed categories into which they were put.
As such we know that dharmas were all impermanent, unsatisfactory and insubstantial. I've already outlined the Sarvāstivādin argument for the "existence" of dharmas in the part and future as well as the present. Arguably one could maintain this kind of view without insisting that dharmas are absolutely real. The first step on the road to seeing dharmas as real seems to have been the fixed categories into which they were put. Whereas in the early Buddhist worldview everything was process, the introduction of fixed categories introduced an artificial reference point into the picture. A dharma was a member of a category in an absolute sense (pāramārthika) and thus a dharma had an identity (ātmabhāva) which was not contingent (Cox 560). It was not inevitable that Buddhists would come to think of their categories as territory rather than map, but it was a slippery slope.

We suffer a similar problem today. For example physicists who study the regularity and similarity of experience can produce highly sophisticated mathematical models which describe the motion of bodies at difference scales and they may or may not be naive realists who believe they are describing reality (in fact most distinguish the map from the territory). But a few tiers down, those doing undergraduate physics are more likely to be unsophisticated about the distinction and to be naive realists. I know this from experience, for as one reproduces the results of, say, Isaac Newton and derives the laws of motion from first principles, the compelling conclusion is that one is describing reality (no one who has not done this can really understand how compelling it is). Once the information goes through the hands of science journalists and into the general public, most of the potential sophistication is lost. Most of us are naive realists with respect to experience, even when we intellectually espouse various philosophies about ontology, in practice we feel ourselves to be in contact with reality.

It is all too easy for human categories to start to seem natural.
Sectarian Buddhist intellectuals were still pre-scientific and produced a variety of speculative views about the world of experience with varying degrees of realism and idealism. In such a milieu the critique of views was simply a clash of opinions. It is all too easy for human categories to start to seem natural, that is to be an aspect of the world rather than something we impose on the world to help us make sense of it.

Think of categories like large and small. These are defined on the basis of prototypes of various kinds of item. When we mention, say, 'dog' each of us has a prototype image of what a representative dog looks like. From this we know that great-danes are large dogs, and chihuahuas are small dogs. We don't usually stop to wonder why, or even pause to apply the label for extreme cases like these. To us it just seems natural. And so, mostly like, the Ābhidharmikas began to think of their artificial categories as natural and therefore real. They begin to blur the distinction between map and territory because the map is an internal, almost a priori construct, that transparently structures the way we interpret experience. Thus began the reification of dharmas by Ābhidharmikas.

Putting Dharmas Into Categories

Colette Cox, with respect to the Sarvāstivāda, and Noa Ronkin, with respect to the Theravāda, both argue that ontological thinking is not obvious well into the Abhidharma project - into what Cox calls the mid-period texts. However dharmas were thought to fit into categories by virtue of their svabhāva - which early on means their 'particular nature'. It is the svabhāva of the dharma that gives it a particular quality and/or function and allows us to categorise it. At first it is simply that we are able to perceive certain regularities in our experience and thus to conceive of different kinds of dharmas. There is something particular about the the experiences that we learn to recognise and give a name to.

By fixing the definition of categories one almost cannot avoid fixing the members of the category.
The problem is that when the category is fixed and hard-edged, then the quality which qualifies any item for membership in that category can also come to seem fixed. Just as categories that move around and are not fixed don't make for very useful taxonomies, it is also practically unhelpful if the members of the categories are able to move around at random. By fixing the definition of categories one almost cannot avoid fixing the members of the category. And thus the meaning of the word svabhāva drifts from 'particular nature' towards 'intrinsic existence' and dharmas start to become real entities (dravya). 


It is the inherent dynamics of encyclopedic projects that, almost inevitably it seems, causes those who pursue such a project to first see their categories as real, and then to see the distinctions they make to fit objects into their categories as real, and in the Buddhist case to see the members of categories, i.e. dharmas, as real. despite the fact that the real/unreal distinction is specifically said to be unhelpful by early Buddhist texts. Be that as it may, the Abhidharma project has had a massive influence on Buddhism and leaves us with a legacy of ontological thought that frequently obscures the true intent of pratītyasamutpāda, i.e. explaining the arising and passing away of mental processes.

It may be that this is why the anti-Realist and anti-Idealist ideas epitomised in the Kaccānagotta Sutta feature in the first chapter of the Aṣṭasāhasrika Prajñāpāramitā Sūtra as well as in Nāgārjuna's Mūlamadhyamaka Kārikā. The early Prajñāpāramitā sūtras tackle Realism head on with an uncompromising anti-Realism. Where Sarvāstivādins proposed real dharmas, authors of the Prajñāpāramitā sūtras said "no dharmas". The seem to have meant "no real dharmas", but the polemic is phrased in apparently nihilistic terms, presumably for rhetorical purposes. No dharmas. No arising. No ceasing. No pure dharmas. No defiled dharmas. And so on. This approach is sampled and remixed in the Heart Sutra.

However we should not look down on the Sarvāstivādins. No other Buddhist school did much better. Theravāda ontology is scarcely more tenable. The predominant Mahāyāna solution to the problem of action at a temporal distance (the Yogacāra ālayavijñāna) involved the invention of metaphysical entities that only disguised the problem. While the Abhidharma was very intellectually productive, it was ultimately a dead end in terms of practice and soteriology.

This essay started by arguing that understanding the Sarvāstivāda was important for seeing Prajñāpāramitā in context. However it also highlights that we are not yet able to say how Prajñāpāramitā deals with the problem of Action at a Temporal Distance. The solution widely adopted by Mahāyāna schools is from the Yogacāra, but there were several centuries between the composition of the Prajñāpāramitā and Yogacāra texts. Nāgārjuna's proposed solution is to treat the everything as an illusion (!) which seems an even less successful answer than the pudgala. I hope in the near future to explore if and how the early Prajñāpāramitā dealt with Action at a Temporal Distance.

We've always known that Buddhism splintered into sects and that the sects had different doctrines. I hope that this trilogy of essays on the Sarvāstivāda has shed some light on how and why sectarian Buddhist developed in the way it did. The early Buddhists were struggling to make sense of the legacy of confusion in the early texts.



Cox, Collett. (2004) 'From Category to Ontology: The Changing Role of Dharma in Sarvāstivāda Abhidharma.' Journal of Indian Philosophy 32: 543-597.
Lakoff, George (1990). Women, Fire and Dangerous Things:  What Categories Reveal About the Mind. University of Chicago Press.
Ronkin, Noa. (2005) Early Buddhist Metaphysics. Routledge.

Related Posts with Thumbnails