Nāgārjuna on Emptiness A Comprehensive Critique of Foundationalism
by Jan Westerhoff
Nāgārjuna is one of the most influential philosophers of ancient India, yet we know very little about him as a person. There is no certainty about when and where he lived, what he wrote, and even about how many Nāgārjunas there were in the first place. Plausible estimates date Nāgārjuna somewhere in the first three centuries ce; there is some suggestion that he was born in South India.1 Apart from the Mūlamadhyamakakārikā, which are attributed to Nāgārjuna in the same way the Iliad and Odyssey are attributed to Homer, and a set of five or six additional works that are commonly regarded as authentic, a large set other of works probably composed over the span of several centuries are ascribed to Nāgārjuna by the various Buddhist traditions.2 Hagiographical accounts ascribe to Nāgārjuna an extraordinarily long lifespan, which is used account for this diversity of works. Modern Buddhology has instead opted for a multiplicity of up to four different authors named “Nāgārjuna” who composed these works over a large stretch of time. Our present discussion will be based on the philosophical discussion found in the Mūlamadhyamakakārikā together with a set of key philosophical works of fairly plausible authenticity.3
The Madhyamaka Project
If we consider the historical development of the Buddhist philosophical tradition founded by Nāgārjuna, Madhyamaka occupies a middle position, arising later than the Abhidharma, but earlier than the Yogācāra.4 Its project is closely connected
with the explication and argumentative support of positions described in a set of Mahāyāna sutras called “Perfection of Wisdom” (prajñāpāramitā) texts, which probably arose in around the beginning of the first millennium ce. The historical influence of Madhyamaka has been considerable. During the centuries after Nāgārjuna it challenged its greatest intra- Buddhist rival, the Yogācāra, in a series of sophisticated debates that are preserved in the Indian Buddhist scholastic literature. After the demise of Buddhism in India in the twelfth century ce Madhyamaka established itself as the dominant philosophical school of Tibetan Buddhism; Madhyamaka continues to flourish in contemporary Tibetan philosophical works. At the beginning of the twentieth century Western scholars of Buddhist studies became increasingly interested in the tenets of the Madhyamaka school. Although much of the traditional commentarial literature still needs to be edited, translated, and analyzed in greater detail, it is fair to say that Madhyamaka is presently the school of Buddhist philosophy most thoroughly studied by Western Buddhologists.
The Different Faces of Non- foundationalism
It is no oversimplification to say that the central point around which Nāgārjuna’s thought revolves is the notion of emptiness. To be empty is, of course, to be empty of something, and the something all things are said to be empty of is what is described by the Sanskrit technical term svabhāva. This term can be rendered into English in different ways: “inherent existence,” “intrinsic nature,” “own- being,” and “substance” are some of the alternatives that have been suggested in contemporary literature. These different translations give us a first idea of the different ways in which the concept of svabhāva can be understood. The term svabhāva is a compound noun, sva meaning “one’s own” or “oneself,” bhāva referring to a thing, to existence, or to a thing’s nature. The core of the concept is the idea that something has this bhāva all on its own, without anything else coming into the picture. For the Ābhidharmikas, the main types of “anything else” are the object’s parts. If an object has parts, the Ābhidharmikas argued, it cannot have its nature in a self- sufficient way, since it only borrows this nature from its parts. It could not exist as a lonely thing in an otherwise empty universe and be still what it is. It is part of the nature of a chariot to convey goods and people, but without the wheels turning and the axle keeping them in place none of this nature could manifest. Some things only are what they are because other things (their parts) are what they are. This continues down the mereological chain until we arrive at the maximally simple objects (the Abhidharma’s dharmas), which have no parts. These things still have natures, but their nature is all contained in them; it does not dissipate into its parts.
It is interesting to note that the Ābhidharmikas did not consider it in any way problematic that their fundamental ontological elements, the dharmas, were caused. While they were concerned that a thing’s intrinsic nature seeps away into its parts, they were not similarly worried about a thing’s causes. That this was so is not entirely surprising given the Abhidharma theories of causation. For a Sarvāstivāda branch of the Abhidharma the fact that a dharma is caused does not really detract from the fact that it has its nature intrinsically, since all objects exist in all three times. What changes when a future object becomes present is that it acquires activity (karitrā), an ability to perform functions that past and future objects do not possess, but its existence does not change. For the Sautrāntika branch each moment of a fire- dharma, for example, is caused by its preceding moment, which is also a fire- dharma. If we thought that a fire- dharma intrinsic nature somehow dissipates into its cause we would still be right to assert that some
fire- dharma (namely the one that is the cause) exists intrinsically. If the only thing that exists is the present moment containing the fire- dharma and if that dharma could exist without anything else being around at the same time, then whether it is caused or not is immaterial for assessing whether it exists by svabhāva. The Mādhyamikas had a fundamentally different understanding of causation, which led them to deny that a lack of parts made for existence with svabhāva and which entailed that they understood the notion of svabhāva differently from the Ābhidharmikas. For the Mādhyamika causation essentially involves the conceptualizing mind, due to the fact that either the cause or the effect fails to exist at the present moment, and thus has to be supplied by the mind. And if this is the case then caused things do in fact borrow their nature from other things. An atomic, causally produced dharma could not have existed in an otherwise empty universe, since there could not have been a causal relation without the conceptualizing mind supplying either the antecedent by memory or the consequent by anticipation. For this reason Nāgārjuna conceptualized existence by svabhāva in a more stringent way. Atomicity was not sufficient; a thing also had to exist without causes and conditions.5 What this boils down to is an understanding of existence by svabhāva in terms of the absence of dependence relations. An object that has parts depends for its existence on the simultaneous existence of these: if there are no blocks in the Lego structure, there is no Lego structure. Conversely, something genuinely atomic exists by svabhāva in this mereological sense, simply because it exists and there aren’t any parts it could depend on. Of course things can also depend on things that no longer exist, as when something depends on its causes. The oak tree could not exist without the acorn, even though the acorn is past and
no longer exists. An uncaused or self- caused thing would exist by svabhāva in this causal sense. Of course something can exist by svabhāva in the mereological sense and fail to do so in a causal sense. The foundational elements of the Abhidharma ontology, the dharmas, do not have parts, yet they are causally produced.6 Even mereologically and causally independent things may still be considered as dependent if we consider the notion of conceptualization- dependence. If we cannot
conceive of a thing’s existence without bringing in a conceptualizing mind (a position that the Mādhyamika accepts) then even an atomic, uncaused thing may fail to exist by svabhāva, because it existentially depends on a mind that underwrites its existence. In this way the conception of emptiness we end up with is an ascending hierarchy of ever more subtle conceptions of what it means to exist without svabhāva. At the coarsest level, all this means is for something to have parts, a view that the Mādhyamika equates with the Abhidharma position. We might be puzzled as to why the Ābhidharmikas would consider this weakest sense of emptiness as exhaustive if we don’t take into account that this position looks considerably stronger if we pair it with an Abhidharma, rather than a Madhyamaka, understanding of causation. Ascending the conceptual hierarchy we realize that there are even more subtle ways in which one thing can derive its nature from another: from its cause via the causal relation, or from a conceptualizing mind. All of these three kinds of dependence relation (mereological, causal, or conceptual) allow the svabhāva of a thing to dissipate into the dependee and thus to disappear. Since later Madhyamaka commentators argue that nothing is independent in all three ways (though things may still be independent in one of the two less subtle ways) at the final level of analysis nothing exists by svabhāva.
Ontological Non- foundationalism A central argument the Mādhyamika uses to establish the emptiness of all things relies on the two concepts of causality and momentariness.7 The idea of momentariness in particular is crucial. This is the view that all things are momentary, which means that they last only for a temporally non- extended instant and immediately self- destruct after that instant, causing, as they do, another, temporally succeeding thing that is very much like, though distinct from, the first thing. This conception is then combined with a presentist view of existence: neither the past nor the future, but only the present exists. Both positions, momentariness and presentism, are of course far from uncontroversial. The Abhidharma presents various arguments in their support; considerations of brevity do not allow us to go into them here in detail. We will therefore just accept them as primitive.8 The second ingredient is the Mādhyamika conception of causation. It is plausible to assume that causation is generally not simultaneous, but that cause and effect are successive. This means that in a world with a temporally minimally extended present only one member of the causal relation can happen during that present, and hence be existent. This leads to an immediate problem with causation conceived of as a two- place relation: we cannot have a relation between two objects if one of them fails to exist. The way Bhāviveka’s commentary suggests we should solve this problem is by thinking of a way in which cause and effect can be made to exist both in the present. The idea is that when the cause exists, the effect is supplied by anticipation, and when the effect exists, the cause is supplied by memory. In this way we can speak of a causal relation in a momentary, presentist world. However, this implies that causation is essentially mind- dependent;
if there were no minds, there could not have been a causal relation. Bhāviveka can then use this result in order to argue that anything causally produced must be empty. For if it is part of the nature of some object to be causally produced, and if causal production involves minds as an indispensible part, then it neither can exist substantially in the most subtle of the three ways (being independent of conceptualization), nor can it bear its nature intrinsically, because being what it is presupposes the existence of mental states distinct from it. This argument is subject to two limitations. First, it relies on the contentious theories of momentariness and presentism, which need to be independently established. Second, it does not amount to an argument for the emptiness of absolutely everything, but only covers causally produced things. The greater the amount of non- caused objects in your preferred ontology, the less convincing the argument will be. Depending on your view of the arguments for momentariness and presentism the first point may or may not be a cause of concern. The second would not have appeared too problematic to an ancient Indian audience, as the Indian philosophical context traditionally did not give much
prominence to abstract objects.9 When studying Nāgārjuna’s works and the later commentaries on them it becomes apparent that there is no argument that is regarded as a “master argument” for universal emptiness, no argument that is considered to be the final (and presumably conclusive) proof that all things lack svabhāva. The reasons for this are intricately connected with Madhyamaka’s view of its own function. Nāgārjuna makes it clear that he regards his account not as a final philosophical theory to replace all others, but as a medicine administered to those who are suffering from the illusion of substantial existence. As the causes of illness are manifold and as a medicine has to be put together to remove the particular cause in question, the argumentative reasons why people believe that items with svabhāva exist vary. For this reason there cannot be a single
argument for emptiness that acts as the philosophical equivalent of a panacea. Rather, arguments have to be tailored to fit the specific case. The argument we just considered is directed against an opponent who believes that causally produced entities can still exist substantially. I would like to consider one more argument against ontological foundationalism here, one directed against the idea that there are objects that constitute the most basic level of reality. There are various ways in which these objects (or dharmas, as the Abhidharma calls them) can be conceptualized in ontological terms. The most straightforward is in terms of the familiar pin- cushion model: each fundamental object is an individual (the pin- cushion) characterized by various fundamental properties (the pins) that are attached to it. The fundamental objects of the Abhidharma include the “four great elements” of earth, water, fire, and air, together with their properties. The difficulty with conceiving of these elements according to the pin- cushion model is that once
we mentally subtract their properties there is nothing left: without its heat, there is no fire; without its wetness, there is not water. This leaves us with the notion of a bare particular. A bare particular is an individual from which all properties have been abstracted away, it is a mere anchor to which a variety of properties can attach so as to be considered properties of a single object. From the perspective of a proponent of svabhāva, however, the introduction of
bare particulars might not be a prudent choice. Suppose the bare particular we have just abstracted from a water atom has its nature intrinsically, independent of anything else, thus existing by svabhāva. Since nothing can have two svabhāvas, wetness would no longer be the svabhāva of water, solidity would not be the svabhāva of earth et cetera, contrary to what we have just assumed. Suppose then the bare particular did not have its nature intrinsically, but was dependent on something else. This is hardly satisfactory, since the whole point of the theory of fundamentally real entities we are considering here is to arrive at something that does not itself depend on anything else for its existence. If the water atom depends on its bare particular, and if that in turn depends on something else, we seem to have contradicted our initial assumption that the objects under discussion are fundamentally real. Since the problem replicates if we analyze the underlying level of reality according to the pin- cushion model as well, we are headed for an infinite regress. Analyzing the idea of a fundamental object using the pin- cushion model brings with it the notion of a bare particular, and this either makes the individual characteristics of these objects superfluous, since it alone is sufficient for establishing existence by svabhāva, or it leads us into an infinite regress. Furthermore, the notion of a bare
particular would certainly be a peculiar choice for someone pursuing a theory of ultimately real fundamental entities. According to the view advanced by the Mādhyamika’s opponent such objects would also be considered as mind- independent. But bare particulars are hardly the first kinds of things that come to mind in this context. We cannot see, feel, or touch such objects; they seem to be obvious examples of explanatory posits of a philosophical theory. To bring in such highly conceptualization- dependent entities as part of a theory that is supposed to account for what is conceptualization- independent is at least somewhat disingenuous. But, the opponent may object at this state, perhaps the problems encountered do not result from the assumption of ontologically fundamental entities, but from a wrong way of conceiving of such entities. The difficulties connected with the
notion of a bare particular spring from the pin- cushion model of individuals, so perhaps it is this model we should discard? An alternative model we could employ here is one that regards the most fundamental entities as particularized properties (or tropes, to use the technical philosophical term). Such tropes differ from repeatable properties in being spatiotemporally individuated. For the trope theorist there is not one property of redness shared by all the British post- boxes, but the spatiotemporally individuated red- trope of the post- box over here, which is entirely distinct from the similarly individuated red- trope of the post- box over there. The advantage of this account is that we do not need a two- category ontology, with all the qualities in one category, and whatever has the qualities (the individual, or the bare particular in the case above) in the other. Rather, what we usually conceive of as an individual is just a conglomeration or co- presence of tropes without an underlying metaphysical condensation nucleus. The postbox is just the
co- presence of a red- trope, a shape- trope, a weight- trope, and so forth. A question that immediately arises in this context is how the different tropes are going to be told apart. The usual answer is that this is done by reference to their different spatiotemporal locations; since each such location is unique all instances of a particular
quality at that location can be distinguished. However, this solution just postpones the problem. For presumably “being at a certain place” and “existing at a certain time” are properties as well, and if a basic assumption of our ontology is that all that exists are tropes, these have to be considered as place- tropes and time- tropes as well. In this case redness at a particular space- time turns out to be in fact a co- presence of tropes (now involving temporal and spatial tropes), so we are back to our original problem. The best way of solving this issue is to individuate tropes in terms of the other tropes they are co- present with. We could then say that trope a is different from trope b because a is co- present with tropes u, v, w, while b is co- present with tropes x, y, z. This procedure allows us to individuate tropes, but it is inconsistent with our initial assumption
that tropes exist by svabhāva. If tropes are the kinds of things that do not borrow their nature from other kinds of things it is not clear how they could at the same time be dependent for their identity on other tropes. If a given trope only is what it is relative to a complex of co- presence it cannot be what it is all on its own. We have now seen how two popular models for understanding what kind of thing the most fundamental sorts of objects could be may be analyzed from Nāgārjuna’s perspective. Both the pin- cushion model and the trope model have been found wanting, as both are affected by significant problems. This may give rise to the suspicion that it is not actually the models that are at fault, but the underlying assumptions that there are fundamental ontological building- blocks that exist with svabhāva. Of course this argument, like all proofs by cases not built on a clearly determined set of alternatives, is not conclusive. The opponent might come up with a better ontological theory of the basic dharmas that does not face any of these difficulties. But before this is spelt out in detail this can hardly be considered as a defense of the opponent’s position. As we noted above, the Nāgārjuna’s aim is not to come
up with a proof that establishes emptiness once and for all, but rather to show for each foundationalist theory the opponent puts forward that it faces fundamental problems. We will therefore now turn our discussion to some other areas of philosophy where foundationalist accounts may be defended. Epistemological Non- foundationalism One such area is epistemology. Epistemological foundationalism is here understood as the claim that there are certain epistemic instruments (such as perception, inference, analogy, or testimony) that by their very nature transmit knowledge. In order to establish the credentials of some claim, all we need to do is to ascertain whether it has come to us through an acceptable epistemic means and whether this means has been correctly applied (i.e., whether the perception is free from error, the inference is valid, and so on). Understood in this way, the epistemic instruments provide a way out of Agrippa’s trilemma, which argues that any knowledge claim must either be based on axioms (and therefore ultimately without justification), or depend on other claims that in turn require justification, going all the way down (leading to an infinite regress) or all the way round (leading to a vicious circle). If I can justify that I know
there is a teacup in front of me by referring to the fact that I see it with unimpeded vision I seem to need to accept no more than that there are certain procedures that by their intrinsic nature lead to knowledge. If I don’t make internalist assumptions I do not even need to know why or how these procedures do so to be justified in my relying on them. As we have just seen the theory of emptiness lends no support to belief in objects with intrinsic nature, and in fact Nāgārjuna subjects the theory of epistemic instruments and the epistemic objects they cognize (pramāХa and prameya) to a sustained critique. He begins by considering the question of whether it is the epistemic instruments that establish the epistemic objects, or, conversely, whether the epistemic objects establish the epistemic instruments. Any defender of a theory of epistemic instruments is obviously going to claim
that they establish their objects. It is, after all, through these instruments that we get to know the objects, and without the instruments there would be nothing known, and therefore no epistemic objects. But Nāgārjuna argues that the establishment also goes the other way round, from epistemic objects to epistemic instruments. Obviously, if there were no epistemic objects, there would be nothing called “epistemic instruments.” But the dependence of epistemic instruments on epistemic objects goes deeper than mere notional dependence; the dependence at issue here is constitutive. The property of being an epistemic instrument is dispositional: when an epistemic object of the right kind comes along, the instruments make the object known to us. This shows us that the very nature of an epistemic instrument cannot be made sense of without that of an epistemic object. Consider the following example. Suppose I present you with a sample of liquid and ask you to determine whether it is a glue or not. A detailed analysis of its chemical structure will not
help you to answer this question, since there is no single structure responsible for all glues’ glue- ness. Alternatively, you could test the substance on various substances, such as paper, marble, styrofoam, et cetera, to determine whether it glues them to each other. Unfortunately this procedure is inevitably piecemeal. Even if the substance proves to be inert with reference to all the substances, it could still be the case that it acted as a very powerful glue on ones you had not yet tested, or even on new kinds of artificial substances that have never been manufactured so far. What this shows is that being a glue is not an intrinsic property: it is not a quality a substance can have solely by itself, but a property that constitutitvely depends, and hence essentially refers to objects it can operate on, making them stick together. The same, the Mādhyamika will argue, is true of epistemic instruments.10 Something can only have the dispositional property of being able to make other things cognitively accessible in dependence on the presence of these
things. If there are no knowable things, it is hard to conceive what it would mean for something to be able to make things known, in the same way as it is difficult to conceive of glue in a universe devoid of things to be glued. This demonstrates that epistemic instruments and the epistemic objects they give us access to are related to each other by mutual dependence. There cannot be one without the other, and as such the property of being either cannot be understood as being possessed by anything intrinsically.
It is important to be aware of the purpose of Nāgārjuna’s argument against epistemological foundationalism. It is directed against an opponent who claims that there must be epistemic instruments that are such instruments intrinsically, since we know things, and we can only do so by employing instruments of this kind. Nāgārjuna then points out problems for this understanding of epistemic instruments. He does so not in order to infer that, because there are no intrinsic epistemic instruments, nobody knows anything at all, but to argue that we have to conceive of our route to knowledge in a different way: epistemic instruments without svabhāva are able to deliver knowledge of the world to us
without the problematic foundationalist assumptions. The fact that we can only make sense of anything being an epistemic instrument in relation to something else (namely an epistemic object) does not mean that we cannot distinguish between those instruments that provide reliable routes to knowledge and those that do not, just like the fact that we cannot look into a substance to see whether it is a glue does not entail that nothing is a glue, or that everything is. This point links up with the absence of a master argument for emptiness. The reason that we find a variety of different arguments for emptiness in the Madhyamaka literature is not that its authors found none of them completely convincing, but that since they all fail to be intrinsically knowledge- producing, different contexts and different situations require different arguments for emptiness. In fact such a situation is just what we would expect if no epistemic instrument can be considered to deliver its result independent of what is going on around it. We should note, however, that Nāgārjuna’s critique of intrinsically existent epistemic instruments and objects is not an epistemic argument for emptiness. One could imagine that someone made an argument along the following lines. The defender of existence by svabhāva, it may be argued, has to uphold the distinction between ontology (what there is) and epistemology (how we find out about it) as fundamental. But Nāgārjuna’s arguments threaten that distinction. Since we can never know whether something is a reliable epistemic instrument without taking into account the epistemic objects at the same time, ontology and epistemology become inextricably linked. Therefore there cannot be a fundamental distinction between them, and thus we cannot have one side of the divide being characterized by objects existing in the way they do intrinsically, by svabhāva. The difficulty with this argument is the
internalist assumption it makes. An internalist assumes that we cannot be justified without knowing that we are justified, and that there cannot be intrinsically existent epistemic instruments without us knowing them to exist in this way. But if we drop this internalist assumption there is no prima facie problem with assuming that even though we cannot find out that something is a good instrument for cognitive access to the world without looking at what objects it provides access to, this does not mean that its being such an instrument is similarly dependent. Doing without the internalist assumption is preferable if we want to understand the Mādhyamikas in their historical context, for their epistemological arguments are generally directed against the Naiyāyikas, who did not hold such internalist assumptions. It is therefore most convincing to regard Nāgārjuna’s critique of epistemological foundationalism as a response to a specific criticism (that there have to be intrinsically existent epistemic instruments), rather than as a stand- alone argument for emptiness.
102 Jan Westerhoff Linguistic Non- foundationalism When considering linguistic foundationalism we have to consider two different varieties of this view, one of which is likely to strike us as less plausible than the other. The first is the idea that there is an intrinsic connection between expressions of a given language and their referents. This idea goes back to the ancient Indian philosophical school of Mīmāьsā, whose adherents believed in an intrinsic connection between (Sanskrit) expressions and the objects they denote. We do not need to refer to specific Madhyamaka arguments to point out the implausibility of this position. The second position, ultimately going back to the Nyāya school, merits more
detailed systematic attention in the present context. Here the idea is not that a Sanskrit word like go by its very nature picks out a cow, but rather that the place the noun cow occupies in the structure of English (or the place go occupies in the structure of Sanskrit) corresponds objectively to the place cows occupy in the structure of reality. The idea behind this so widespread, and so intuitively convincing, that we might call it the “standard picture” of semantics. According to this picture the world out there is objectively carved up “at the joints” into various individuals, their properties, and relations between them. All languages that successfully speak about the world share the same structure, and in the case of a true sentence its structure is the very same as the structure of the piece of the world it talks about. This view is a form of foundationalism since matters of reference and truth are ultimately grounded in the structural isomorphism between expressions and expressed. What are Nāgārjuna’s reasons for rejecting the standard picture?
This picture has the notion of svabhāva written all over it. The world has the structure it has intrinsically, and the greatest part of its division into individuals and properties is mind- independent. The language we use about the world is of course a human creation, but it manages to speak about the world because it intrinsically has a structure that is identical with the intrinsic structure of the world. When discussing language Nāgārjuna mentions one argument to the effect that words and their referents are neither identical nor distinct by their very nature, by svabhāva. If they were identical we would end up with absurd consequences such as the word “fire” burning our mouth; if
they were distinct and wholly unrelated we would find it hard to explain how we can use one to refer to the other. The standard picture seems to offer a way around this difficulty: our language refers to the world via a structural isomorphism; this structure connects them without being identical with either. Yet the matter might not be as straightforward as this. Abstract objects (such as structural isomorphisms) might connect world and word, yet these objects are also things we can talk about. So to explain reference to them, we appear to need another structure that our language and the first structure both exemplify, and to talk about this we need yet another one. When trying to spell out fully what the fundamental structural equivalence connecting language and reference is we seem to be headed for an infinite regress.
What, then, is the alternative picture of language the Nāgārjuna suggests instead of the standard picture? It is one in which the connection between objects and language is based on conventions, and, more importantly, one in which there is also no preexistent, foundational division of the world “by the joints” into a set of individuals, properties, and relations. The division of the world into things is as much convention- based as the set of linguistic items we use to refer to them. Nāgārjuna makes this point quite explicit when he points out that “the whole world is nothing but words” (nāmatra jagat sarvam).11 It is important to be aware, though, that what he has in mind here cannot be the kind of cookie- cutter approach to conceptualizing we sometimes find in modern accounts of anti- realism.12 There the idea is that our conceptual schemes work like cookie- cutters applied
to the world; different cookie- cutters cut out different shapes, but there is no set of preexisting cookies already in the dough. It is not, in other words, already “cut at the joints.” Our conventions correspond to the cookie- cutters, and these determine what individuals and what objects there are. The difficulty with this theory is that it commits us to the view of a preexistent cookie dough that is there independent of and prior to the applications of our concepts. We may shape the world by our minds, but it is there independent of any shaping. Nāgārjuna’s theory, by contrast, is considerably more radical. He considers not just the form of the world around us to be convention- dependent, but its very existence. How to spell this theory out in detail is not entirely straightforward. We have no difficulty accepting that the existence of some objects is wholly convention- dependent (items traded in the stock market, for example), but these appear to be significantly different from things like rocks, tables, and stars. One way of trying to make the basic idea more precise is by arguing that the objects around us do not exist, but that they are just constructions from mental events, which are entirely subjective and only accessible by the mind that holds them. We may then claim that external objects just reduce to certain convention- based ways of talking about such subjective events.13 In this way no objectively existent world, no cookie- dough, would have to be assumed as a basis that is cut up by conventionally established concepts. Rather, the establishment of conventions establishes the very objects that form part of what appears to us as a convention- independent world.
Non- foundationalism about Truth and the Problem of Nihilism One of the most noteworthy aspects of Nāgārjuna’s non- foundationalism is that it is not confined to specific (if fairly comprehensive) topics such as ontology, epistemology, or language, but includes a very general non- foundationalism about truth as well, a non- foundationalism that also applies to the theory of emptiness itself. For the foundationalist about truth, the world is fundamentally and objectively in a certain way, and the aim of a good philosophical theory is to capture this way. Foundationalism about truth is independent of our views on whether or not the ultimate truth about the world is epistemically accessible to us. A skeptic who claims that we can never know what the world
is really like at the fundamental level can still be a foundationalist about truth insofar as he believes that the world is ultimately some way or another. In this context it is interesting to consider a dilemma that has been brought forward against Nāgārjuna’s theory of emptiness14 and forms of non- realism more generally.15 One horn of the dilemma is that the Nāgārjuna’s own theory is an exception to his non- foundationalism about truth. In this case the theory of emptiness is underwritten by the way the world is at the most fundamental level, but this simultaneously implies that there is that least one exception to Nāgārjuna’s claim that truths in general are not based on the way reality is ultimately. The second horn is that Nāgārjuna’s theory is not underwritten in this way. In this case the theory seems to saw off the philosophical branch it is sitting on. For if
it is not a theory of what the world is like at the most basic ontological level, what kind of theory is it in the first place? Nāgārjuna’s non- foundationalism appears to leave us only with the choice between inconsistency and vacuity. As is evident from their embracing the idea of the “emptiness of emptiness,” Nāgārjuna accepts the second horn. He points out that its theory only appears philosophically impotent if it is set up in contrast to accounts that accept foundationalism about truth and that propose to expound what the nature of the world is ultimately. In this case Nāgārjuna’s account seems about as impotent as rain from a meteorological simulation when compared to real rain. However, it is precisely this contrast that Nāgārjuna rejects. His arguments set out to show that the idea of a theory that represents the nature of reality at the ultimate level is a chimera. Any such theory will entail the existence of some entities by svabhāva, and any such entities, Nāgārjuna claims, will lead to inconsistency. For this reason embracing the second
horn of the dilemma and accepting that his own theory of emptiness is as empty as anything else is not regarded by him as being in any way detrimental to the theory of emptiness. The Problem of Nihilism Madhyamaka has frequently been characterized as nihilism; these claims differ in the acuity of their understanding of the Madhyamaka project and thereby in their philosophical plausibility. One of the more interesting points made in this context (already raised by Asaюga) is that a non- well- founded dependence hierarchy that descends infinitely is never able to trace the chain of dependencies back to something non- dependent. And, the objector argues, if everything depends on something else, and this will in turn depend on something else different from it, existence will dissipate in the unending hierarchy of levels. There are two main ways in which the Mādhyamika can reply to this charge, either by a qualified acceptance of the nihilist conclusion or by its rejection. In the first case the Mādhyamika accepts that the assumption of an infinitely descending hierarchy of levels, together with the belief that only the fundamental really exists, leads to the conclusion that nothing really exists. Yet this view is neither contradictory nor
absurd, since it does not entail a rejection of the variety of appearances that make up the world, only of the claim that such appearances entail the existence of something existing substantially. None of such fundamental concepts like causation, moral value, and karmic consequences have to be given up, but, being relegated to the level of mere appearance, they are not given any fundamentally real status. Alternatively the Mādhyamika can reject the idea that an infinitely descending dependence hierarchy leads to a dissipation of existence. The first step in doing do so is to point out that even though a Mādhyamika is committed to reductionism, he need not be an eliminativist. While an eliminativist will deny the existence of some xs that depend on some ys and claim that the only thing that does in fact exist are the ys, a reductionist can still ascribe some
limited sense of existence to the xs. They exist insofar as they are practically useful objects of reference, and even though they do not share the existential status of the fundamentally real ys, they exist in a manner of speaking. For mereological reductionist wholes, being wholly existentially dependent on the parts that compose them, are not included in “what there is” on the basic level. Yet he does not want to assign wholes to the same realm as golden mountains, square circles, and horns of rabbits, which, unlike wholes, have no role to play in the daily discourse that helps us to navigate the world successfully. Nevertheless, the opponent will object, all this shows is that the Mādhyamika would be justified to ascribe some limited existential status of objects that lack svabhāva only if their reduction- base, the objects they fundamentally depend on, exists by svabhāva. But since Nāgārjuna rejects that anything exists in this way, it cannot conceive of itself along the lines of the reductionist enterprise. At this point the discussion boils down to the question of whether we can consistently conceive of non- well- founded hierarchies of existential dependence. Nāgārjuna rejects the existence of any endpoints of
such a chain, that is, objects on which other objects depend, but which do not in turn depend on anything. The only two structures lacking such objects are those where every chain of dependence is either infinitely descending or loops back onto itself. Are such structures possible? When we are concerned only with question of consistency it can be shown that such structures do not entail hidden contradictions. Both can be modeled in a variety of set theory called anti- founded set theory, and this theory is demonstrably consistent.16 Of course this is no proof of the correctness of the theory of emptiness (there are plenty of consistent structures that fail to be identified in our world), and the opponent might well object that the idea of a downward infinite or loopy dependence structure is unsatisfactory for reasons other than merely logical ones. Nevertheless, it is important to realize that the criticism that the defender of the theory of emptiness is trying to do something that is inconsistent by its very nature (a charge that we find both in historical17 and contemporary18 criticisms of Madhyamaka) is unfounded. The question of the viability of Nāgārjuna’s ontological vision will have to be settled by more general philosophical considerations and cannot be made to rely on the question of consistency alone.
The influence of Nāgārjuna’s Madhyamaka on the development of Buddhist thought (as well as its influence on Indian philosophy more generally) can hardly be overestimated. Mādhyamikas engaged in debates with other Buddhist and non- Buddhist schools right up to the disappearance of Buddhism in India in the twelfth century ce. The elaboration of Madhyamaka philosophy continues up to the present day in an unbroken succession through Tibetan Buddhism. Nāgārjuna’s thought (as interpreted through Candrakīrti and Buddhapālita) became the foundation of the dominant philosophical school in Tibet and is still regarded as the pinnacle of theoretical sophistication in the Tibetan tradition. The sustained
philosophical engagement with Nāgārjuna’s thought in the West goes back about to the beginning of the twentieth century19 and was significantly characterized by the philosophical fashions of the time. A phase of reading Nāgārjuna’s philosophy as a form of Kantianism was followed by one in which the tools of logical formalization were unleashed on the Mūlamadhyamakakārikā; subsequently there was a period in which Nāgārjuna was studied through a post- Wittgensteinian lens.20 While this should caution us against projecting the problems and questions of the day back into ancient Indian discussions there is, nevertheless, a significant connection between the Madhyamaka enterprise and the discussion of ontological dependence, grounding, and foundation popular in the metaphysical discussion of this first decade of the twenty- first century. This is because Nāgārjuna provides an ancient, influential, and consequently well- developed attempt to take the non- foundationalist agenda as far as it can go. The focus of the theory of emptiness in early Buddhism was the idea of the non- substantiality of the person or self; it then developed more and more from a local ontological thesis concerning one specific entity to a general philosophical thesis covering all things. In the hands of the Madhyamaka the theory of emptiness was developed into a variety of possible directions, covering epistemology,21
language and logic,22 ethics,23 and, in a final metaphilosophical turn, the theory of emptiness itself. Madhyamaka therefore presents a fascinating sample case not only for answering the question of how global one’s non- foundationalism can be, but also concerning the consequences of a rejection of foundationalism for the philosophical enterprise itself. If we are non- foundationalists about truth, the philosophical theory we end up with obviously cannot be regarded as corresponding to the ultimate truth about the universe. But we might then ask ourselves what the point of such apparently metaphysical theories like Nāgārjuna’s theory of emptiness is if they are not trying to tell us what reality is like at rock bottom. Are they caught in a metaphilosophically self- defeating predicament? Or are they offering us a conception of the philosophical enterprise that is radically different from the one we are used to? Because it challenges us to reflect on questions such as these, the study
of Nāgārjuna’s thought continues to be a philosophical research project that bears rich systematic fruit. Notes 1. For a recent attempt at a historical contextualization of Nāgārjuna see Joseph Walser, Nāgārjuna in Context: Mahāyāna Buddhism and Early Indian Culture (New York: Columbia University Press, 2005). 2. A survey of Indian Madhyamaka literature, including the works of Nāgārjuna, is in David Seyfort Ruegg, The Literature of the Madhyamaka School of Philosophy in India (Wiesbaden: Otto Harrassowitz, 1981). 3. This is the so- called Yukti corpus, comprising in addition to the Mūlamadhyamakakārikā, the Śūnyatāsaptati, Yuktitгaгзikā, Vigrahavyāvartanī, VaidalyaprakaraХa, and Ratnāvalī. 4. Whether the school of Diюnāga and Dharmakīrti should be considered as an independent philosophical tradition
distinct from the other three is a matter of debate. 5. MMK 15:2; see also the helpful comments on this passage in Mark Siderits and Shōryū Katsura, Nāgārjuna’s Middle Way (Boston: Wisdom, 2013), 155– 157. 6. The converse, that something uncaused or self- caused has parts, cannot obtain, at least if we assume that mereologically dependent is (synchronously) causally dependent on its parts. In this case causal independence turns out to be a stronger notion than mereological independence. 7. The basis of this argument are Bhāviveka’s comments on Mūlamadhyakakakārikā 1.3 in his Prajñāpradīpa; see Yuichi Kajiyama, “Bhāvaviveka’s Prajñāpradīpaк ( 1. Kapitel),” Wiener Zeitschrift für die Kunde Süd- und Ostasiens und Archiv für indische Philosophie 7 (1963): 37– 62 and for further discussion Mark Siderits,“Causation and Emptiness in Early Madhyamaka,” Journal of Indian
Philosophy 32 (2004): 393– 419. 8. For some further discussion see Mark Siderits, Buddhism as Philosophy (Aldershot: Ashgate, 2007), 119– 123, 130– 137. 9. A modern defense of Madhyamaka would of course have to have something to say on the issue. A conceptualist understanding of mathematical and other abstract objects appears to be a promising avenue to explore here. Abstract objects are then reduced in some way to mental objects, objects to which familiar arguments for emptiness will then more straightforwardly apply. 10. As Nāgārjuna makes clear in Vigrahavyāvartanī 41; see Jan Westerhoff, The Dispeller of Disputes: Nāgārjuna’s Vigrahavyāvartanī (Oxford: Oxford University Press, 2010), 81– 82. 11. Acintyastava 35; Christian Lindtner, Nagarjuniana: Studies in the Writings and Philosophy of Nāgārjuna (Delhi: Motilal Banarsidass, 1987), 152– 153. 12. Hilary Putnam, Realism with a Human Face (Cambridge, MA: Harvard University Press, 1990), 96. 13. For an attempt to work this idea out in more detail see Jan Westerhoff, “The Merely Conventional
Existence of the World,” in Moonshadows: Conventional Truth in Buddhist Philosophy, ed. the Cowherds, 189– 212 (Oxford: Oxford University Press, 2011). 14. In its earliest form by the opponent in the Vigrahavyāvartanī. See Westerhoff, Dispeller.
15. See Paul Boghossian, Fear of Knowledge: Against Relativism and Constructivism (Oxford: Clarendon Press, 2010), 54– 57 . 16. For details of anti- founded set theory see Peter Aczel, Non- well- founded Sets (Stanford: Center for the Study of Language and Information, 1998); for its application to Madhyamaka see Graham Priest, “The Structure of Emptiness,” Philosophy East and West 59 (2009): 467– 480 and One: Being and Investigation into the Unity of Reality and of Its Parts, Including the Singular Object which Is Nothingness (Oxford: Oxford University Press, 2014), 182– 193. 17. Janice Willis, On Knowing Reality (New York: Columbia University Press, 1979), 161– 162. 18. Paul Williams, Altruism and Reality: Studies in the Philosophy of the Bodhicaryāvatāra (Richmond: Curzon Press, 1998), 247, note 86. See also 222, note 26 of that volume, as well as Paul Williams, “Response to Mark Siderits’ Review of Altruism and Reality,” Philosophy East and West 50.3 (2000): 424– 453, especially 439 and Paul Williams and Anthony Tribe: Buddhist Thought (London: Routledge, 2000), 150; David Burton: “Is Madhyamaka Buddhism Really the Middle Way?” Contemporary Buddhism 2 (2001): 177– 190, esp. 181. See also Burton, Emptiness Appraised: A Critical Study of Nāgārjuna’s Philosophy (Richmond: Curzon Press, 1999), 109– 111. 19. Disregarding a short but fascinating encounter between Tibetan and Catholic scholasticism in the 18th century. See Giuseppe Toscano, ed., Ippolito Desideri: Opere Tibetane di Ippolito Desideri (Rome: ISMEO, 1981– 1989).
20. Andrew P. Tuck, Comparative Philosophy and the Philosophy of Scholarship: On the Western Interpretation of Nāgārjuna (Oxford: Oxford University Press, 1990). 21. Westerhoff, Dispeller. 22. Fernando Tola and Carmen Dragonetti, Nāgārjuna’s Refutation of Logic (Nyāya): VaidalyaprakaraХa (Delhi: Motilal Banarsidass, 1995). 23. The Cowherds, Moonpaths: Ethics in the Context of Conventional Truth (Oxford: Oxford University Press, 2015). Bibliography Burton, David. Emptiness Appraised: A Critical Study of Nāgārjuna’s Philosophy. Richmond: Curzon Press, 1999. Burton, David. “Is Madhyamaka Buddhism Really the Middle Way?” Contemporary Buddhism 2 (2001): 177– 190. Cowherds, The. Moonpaths: Ethics in the Context of Conventional Truth. Oxford: Oxford University Press, 2015. Lindtner, Christian. Nagarjuniana: Studies in the Writings and Philosophy of Nāgārjuna. Delhi: Motilal Banarsidass, 1987. Priest, Graham. “The Structure of Emptiness.” Philosophy East and West 59 (2009): 467– 480. Priest, Graham. One: Being and Investigation into the Unity of Reality and of Its Parts, including the Singular Object which Is Nothingness. Oxford: Oxford University Press, 2014. Ruegg, David Seyfort. The Literature of the Madhyamaka School of Philosophy in India. Wiesbaden: Otto Harrassowitz, 1981. Siderits, Mark. “Causation and Emptiness in Early Madhyamaka.” Journal of Indian Philosophy 32 (2004): 393– 419. Siderits, Mark. Buddhism as Philosophy. Aldershot: Ashgate, 2007.
Siderits, Mark, and Shōryū Katsura. Nāgārjuna’s Middle Way. Boston: Wisdom, 2013. Tola, Fernando, and Carmen Dragonetti. Nāgārjuna’s Refutation of Logic (Nyāya): VaidalyaprakaraХa. Delhi: Motilal Banarsidass, 1995. Walser, Joseph. Nāgārjuna in Context: Mahāyāna Buddhism and Early Indian Culture. New York: Columbia University Press, 2005. Westerhoff, Jan. The Dispeller of Disputes: Nāgārjuna’s Vigrahavyāvartanī. Oxford: Oxford University Press, 2010. Westerhoff, Jan. “The Merely Conventional Existence of the World.” In Moonshadows: Conventional Truth in Buddhist Philosophy, edited by the Cowherds, pp. 189– 212. Oxford: Oxford University Press, 2011. Williams, Paul, and Anthony Tribe. Buddhist Thought. London: Routledge, 2000. Willis, Janice. On Knowing Reality. New York: Columbia University Press, 1979.
