In one of his key texts, Mūlamadhyamakakārikā (MMK), Nāgārjuna famously expounds the thesis of emptiness.1 This thesis is the core of the Madhyamaka (‘Middle-Way’) school of Buddhist philosophy, and the MMK is considered to be a foundational text for the Mādhyamikas.2 The thesis can easily be stated: anything which exists is empty of svabhāva. (I will leave the Sanskrit term ‘svabhāva’ untranslated for now.) What the thesis means is a hotly debated question by traditional and modern commentators. One thing that is clear from the discussion of Garfield in his contribution to this volume (Chapter 3) is that Nāgārjuna was not a nihilist. That is, the thesis of emptiness does not mean that nothing exists. Emptiness in Nāgārjuna’s sense is not the lack of everything, or nothing.
In order to work out what it does mean, we first need to examine the notion of svabhāva. Westerhoff (2009, § 2.1) has usefully identified three senses of svabhāva that Nāgārjuna employs in his argumentation (at least according to the sixth-to-seventh century commentator Candrakīrti, whose interpretations were in the margins in India but became very influential in Tibet). First, ‘svabhāva’ is sometimes used to mean essential property. An essential property is what gives an object its numerical identity. It is something that an object cannot lose without becoming something else. For example, if heat is an essential property of fire, something is a fire so long as it possesses heat. In other words, it is heat as an essential property that makes it a fire. If heat were removed from a particular fire, that fire would be extinguished. Second, ‘svabhāva’ is sometimes used to mean independence. In this sense, to have svabhāva is to exist independently of everything else. This means that an object with svabhāva can exist in a possible world that is otherwise unpopulated by any other things, since the existence of an object with svabhāva does not depend on the existence of anything else. To say that an object exists independently of everything else is not just to say that it is the sole occupant of an otherwise empty world, but also to say that it is not composed of anything. If the object with svabhāva has a mereological relation (part-whole relation) to its parts, then it exists in a way that depends on those parts. Svabhāva in the sense of independence also implies that the object is not constituted by things that are more basic or simple. So, an object with svabhāva is unconstructed, not causally based on anything else and not mereologically based on its parts.
If an object with svabhāva exists independently of everything else, any successful analysis of it can’t be done in terms of anything else because svabhāva gives an object an independent existence. Furthermore, it can’t be analyzed away by positing more basic entities since it has no parts and therefore can’t be decomposed any further. Thus, svabhāva is what we are left with when an analysis of a thing reaches its limit. Svabhāva is what we aim to identify as the end point of a thorough analysis and it must not disappear during the process of analysis. The third sense of svabhāva is then that it withstands any (logical) analysis.
Strictly speaking, Nāgārjuna mostly employs the second sense of svabhāva, and we hardly see him using its other senses (Westerhoff 2009, § 2.1). It should be noted in this context that the Ābhidharmikas, against whom Nāgārjuna’s argument is partly directed, argue for a different view of svabhāva. In the Abhidharma literature, which contains an early exploration of Buddhist philosophy based on the teachings of the Buddha, we see arguments to the effect that it is only parts which are real and the whole is unreal. Using the example of the chariot from the popular text Milindapañha (The Questions of King Milinda),3 the relevant ontological view can be presented such that when certain “parts” (a pole, an axle, wheels, the body, the flagstaff, a yoke, reins, and a goad) are put together, there is nothing other than the parts themselves. When there is a collection of parts, that is all there is. There is no entity that is distinct from or even identical with a collection of the parts other than the collection itself. To think otherwise is to think that there is a collection of parts and, in addition, a chariot. Such a view, according to the Ābhidharmikas, is incoherent. In his influential Abhidharmakośa, considered a compendium of Abhidharma literature, Vasubandhu presents this Abhidharma view in terms of particulars which we perceive and universals which we conceive. Putting aside the epistemological dimension of his presentation, Vasubandhu elucidates several mereo-logical arguments for the view that only particulars are real and universals are unreal. According to Abhidharma ontology, there are no universals of which the particulars are instances or which supervene on the particulars.4 When this mereological analysis is applied to the parts themselves, the resulting view is that it is only the simple, partless entities (the ultimate ‘parts’) that really exist. It is these ultimate simple entities that are considered to have svabhāva. But for most Ābhidharmikas, these ultimate simple entities are considered to be causally efficacious. An object with svabhāva arises from other objects with svabhāva based on causation. So svabhāva is, for their theory, dependent on causes. This shows that, for the Ābhidharmikas, svabhāva does not have an independent existence. Nāgārjuna is thus arguing against the notion of svabhāva that can be found in the Abhidharma literature.5
In any case, if svabhāva has the three characteristics described above, then Nāgārjuna’s thesis of emptiness can be restated as follows:
Anything which exists is empty of any essential property that is independent of anything else and that withstands any (logical) analysis.
In the MMK, Nāgārjuna presents several arguments for this thesis. His arguments proceed by showing that we can’t identify the essential and independent nature of anything that exists by means of a thorough (logical) analysis.
After presenting his arguments for the thesis, in Chapter XXIV of the MMK, often considered to be the most important chapter of the text, Nāgārjuna turns his attention to explaining how to understand his thesis of emptiness. He does so by invoking ultimate and conventional realities. If we understand the essential property that is independent of everything and that withstands analyses to give the status of ultimate existence to an object, then the thesis of emptiness implies that nothing ultimately exists. This means that nothing can be subjected to an ultimate analysis that establishes its independent and essential properties. No analysis can show the ultimate existence of anything.
What does exist is explained by Nāgārjuna in terms of conventional existence. What exactly this means is a hard question to answer. However we try to answer it, we must not forget that emptiness is not just an ontological thesis revealing what exists, but also a semantic thesis. The word Nāgārjuna uses to introduce two ‘realities’ as an explanation of emptiness is ‘satya’. In MMK XXIV.8, Nāgārjuna writes:
The various buddhas’ teaching of the Dharma [buddhas’ teaching] relies upon two satye [dual form of satya]: the conventional truth of the world and what is true from the ultimate perspective. (Translated in Cowherds 2011, ch. 3)
The word ‘satya’ can mean ‘real’ or ‘existence’. It can thus be used as an ontological category. But it can also mean ‘truth’.6 Despite the fact that semantics (or language) receives very little attention from Nāgārjuna, we must examine the semantic aspect of emptiness in terms of two truths. As I will show, it is the semantics behind the two truths that is crucial to understanding emptiness.
The notion of the two truths did not originate with Nāgārjuna. We can find the two truths doctrine in Abhidharma literature. According to the Ābhidharmikas, only simple, partless entities exist and everything else, such as chariots, doesn’t exist. Given that simple, partless entities are real, we can talk about them truthfully. Moreover, since those entities are the ultimate reality, a statement about them is ultimately true when it describes them accurately.7 So, for the Ābhidharmikas, there are statements which can be said to be ultimately true.
When we analyze ultimate truth in this way, we are presupposing that there are things that are referenced by the statement. A statement is ultimately true so long as it corresponds to the ultimate way in which the referents exist. A helpful way to understand the semantic principle underlying ultimate truth is to think of it as Russellian (as opposed to Fregean). Early in his career, Russell argued that true propositions (expressed by sentences) were facts, and facts were constituted by objects standing in certain relations to one another. Russell (1904) says thus:
People imagine that if A exists, A is a fact; but really the fact is ‘A’s existence’ or ‘that A exists’. Things of this sort, i.e. ‘that A exists’ … I call propositions, and it is things of this sort that are called facts when they happen to be true. (492)
If this were not the case, we would not know anything (objective) about objects. As Russell (1904) writes in response to Frege:
I believe that in spite of all its snowfields Mont Blanc itself is a component part of what is actually asserted in the proposition ‘Mont Blanc is more than 4,000 metres high’. We do not assert the thought, for this is a private psychological matter: we assert the object of the thought, and this is, to my mind, a certain complex (an objective proposition, one might say) in which Mont Blanc is itself a component part. If we do not admit this, then we get the conclusion that we know nothing at all about Mont Blanc. (169)
Despite the fact that Russell presents his view in an epistemological context, it is not only knowledge about the world but also (true) propositions that get this Russellian treatment. Thus, Russell treats truth purely extensionally. That is, for Russell, truth is only a matter of reference. A consideration of truth is thus a consideration of what exists.
We can similarly understand the nature of the Abhidharma semantic principle behind ultimate truth. Russell can be seen as collapsing the distinction between existence and truth. The word ‘satya’ itself does not make such a distinction. An ultimate truth is thus just what ultimately exists. For the Ābhidharmikas, what ultimately exist are simple, partless entities and nothing else. An ultimate truth is just the (ultimate) way in which these simple, partless entities exist.
In Abhidharma literature, the same semantic principle underlies conventional truths. At least, there is no evidence to suggest that conventional truths are based on a different semantic principle. If an ultimate truth is concerned with what ultimately exists, then a conventional truth is concerned with what conventionally exists. But what are the things that conventionally exist? A chariot can be analyzed further into its parts. So, from an Ābhidharmika’s point of view, a chariot is constructed out of simple, partless entities. But the chariot is unreal, it does not really exist because only its parts really exist. A chariot is thus a “fictional” object. It is just a figment of our imagination. Hence, the statement “a chariot is in the backyard” can’t be an ultimate truth even if its parts can be said to ultimately exist. Nonetheless, uniformity demands that conventional truths are given the same semantic treatment as ultimate truths. Moreover, the Ābhidharmikas do not present a different semantic principle to deal with conventional truths. So, it would seem to follow that a conventional truth is concerned with a conventional reality where chariots and everything else can be said to exist albeit only conventionally.
The semantics Ābhidharmikas adopted were the orthodox account in Indian philosophical circles around the time of Nāgārjuna.8 While arguing against the ontology of Abhidharma, Nāgārjuna does not seem to introduce a new semantic principle. In the context of arguing against the notion of svabhāva, one might say that his semantics is also Russellian. Nāgārjuna does not invoke anything like a Fregean sense that mediates between the truth-bearer and its referent. He, like the Ābhidharmikas, appears to assume a reference-based semantic principle. That is, Nāgārjuna seems to account for the truth of a sentence in terms of reference.
Even though Nāgārjuna’s semantics can be seen to be continuous with previous philosophical developments, it poses some difficulties for understanding his thesis of emptiness. The main difficulty is that ultimate reality is empty for Nāgārjuna. Nothing exists with svabhāva according to his thesis of emptiness. Given that it is svabhāva that bestows ultimate existence, this means that nothing ultimately exists. But if nothing ultimately exists, then there can’t be ultimate truth either, per the semantic principle that Nāgārjuna inherits from the Abhidharma literature. According to Abhidharma semantics, a statement is ultimately true if it corresponds to the ultimate way in which things exist. For Nāgārjuna, however, there is no ultimate way in which things exist and, hence, there is nothing to which a statement can correspond. Hence his semantics, together with his ontology, deliver no ultimate truths; that there are no ultimate truths can’t be an ultimate truth either, since there is nothing to which it can correspond.
The trouble is that it is now unclear what the truth-value of Nāgārjuna’s statements is. Do his statements express ultimate truths or conventional truths? If they express only conventional truths, the Ābhidharmika can reject them because, for them, conventional truths are about fictional entities. If they express ultimate truths, however, his statements are contradictory because Nāgārjuna must also subscribe to the existence of ultimate truths, which is impossible given everything else he says about the emptiness of ultimate truth. How then are we to understand his statements?
It was the contemporary commentators Garfield and Priest (2003) who thematized the paradoxical nature of Nāgārjuna’s thesis of emptiness. After showing that there are no ultimate truths for Nāgārjuna, Garfield and Priest point to a paradox of expressibility. Since there are no ultimate truths, an ultimate truth can’t be expressed. Yet, it is expressed by Nāgārjuna. For example in MMK XXIV.19, he writes:
Something that is not dependently arisen
Such a thing does not exist.
Therefore a non-empty thing
Does not exist. (Quoted in Garfield and Priest 2003, 11)
According to Garfield and Priest, this passage is concerned with ultimate reality, and Nāgārjuna is asserting an ultimate truth. If this is right, then Nāgārjuna is squarely in the realm of paradox.
As Garfield and Priest argue, the paradox Nāgārjuna presents us with is not only about expressibility but also about ontology. To see this, we must remember that Nāgārjuna adopts a reference-based semantics. Under this semantics, to think of his thesis as an ultimate truth is to think that there are things which ultimately exist. But that is exactly what Nāgārjuna sets out to refute. According to Nāgārjuna’s ontology, ultimate reality is empty in the sense that there is no ultimate way in which things exist. So the paradox of expressibility permeates through his ontology by means of his semantics. For Nāgārjuna, ultimate reality is and is not empty, according to Garfield and Priest.
Traditional commentators did not embrace the paradoxes and the contradictions that they entail. The traditional Mādhyamika (and Buddhist) attitude toward contradictions is clear: just reject them. In fact, Nāgārjuna seems to be aware of the contradictory implication of his statements and tries to defuse it. In the Vigrahavyāvaryanī (VV), he says:
If I had some thesis the defect would as a consequence attach to me. But I have no thesis, so this defect is not applicable to me. (Quoted in Westerhoff 2009, 183)
As Westerhoff explains, Nāgārjuna makes this dramatic statement in response to his opponent (real or imagined) who accuses him of being committed to contradictions or ineffectualness. According to the opponent, if his statements are not empty, they contradict his thesis of emptiness; but if his statements are empty, they are ineffectual for refuting the opponent’s view. Nāgārjuna does not consider accepting the first horn of the dilemma. Instead, he accepts the second horn of the dilemma and holds that his statements are empty. He nevertheless rejects the consequent that his statements are ineffectual. In order to do this, Nāgārjuna claims that his opponent’s statements are empty too. He then considers the opponent’s complaint that the alleged ineffectualness of the opponent’s statement is due to Nāgārjuna’s thesis of emptiness. In response to this, Nāgārjuna claims that he has no thesis of his own (Westerhoff 2009, § 9.2).
It is important to note for our purpose that Nāgārjuna’s opponent presupposes a reference-based semantics and assumes that the norms embodied in statements come from references. If Nāgārjuna’s statement of no thesis is effectual as a response to this opponent, one of the things that he needs to articulate is a different semantics. He does not formulate such a semantics even in the VV, however. It is thus not clear how we are to understand his semantics, a semantics that is forced upon him by his opponent.
The Mādhyamika can either replace the orthodox reference-based semantics by another account (the situation Nāgārjuna seems to be forced into) or keep the orthodox account but deflate its problematic consequences.9 For the second option, one can deflate the problematic consequence by adopting a deflationary account of truth. A deflationalist can accept a reference-based semantics so long as it is not metaphysically charged. According to the deflationary account, the statement <s> is true iff s (the T-schema). If we think of s on the right hand side of the biconditional as a fact or a state of affairs, it contains objects that are referenced by the statement <s>. Yet, the referential relation does not have to determine the truth of the statement. All there is to the notion of truth is the T-schema. There is no need to assume that it is the reference relation that guarantees the biconditional. The Prāsaṅgika-Mādhyamika Candrakīrti presents an account that can be described along these lines. He accepts the view that what is true is only “what is acknowledged by the world (lokaprasiddha).” In his Prasannapadā Madhyamakavṛtti, he approvingly cites a passage of the Ratnakūṭa (one of the collections of the Buddha’s teachings):
The world argues with me. I don’t argue with the world. What is agreed upon in the world to exist, I too agree that it exists. What is agreed upon in the world to be nonexistent, I too agree that it does not exist. (Quoted in Cowherds 2011, 151)
Thus, according to Candrakīrti, all we can do in search of truth is simply “read off the surface,” leaving no room for (logical) analysis, as Siderits (1989, 244) puts it.10
Given the small philosophical success Candrakīrti had during his lifetime, it is not clear how widespread his view (or a view like his) was in India. Other Mādhyamika philosophers criticized him for his radical view, however. For example, the eighth-century Svātantrika-Mādhyamika philosopher Kamalaśīla argued against the view of Candrakīrti (or someone like him) by showing that the lokaprasiddha account of truth reduces ‘truth’ to mere beliefs and thus is unreliable (Sarvadharmaniḥsvabhāvasiddhi, p. 312a 8–312b 6).11 Kamalaśīla emphasized the importance of (logical) analysis in accounting for truths because simply referring to people’s opinions is not tantamount to truth. The problem with a deflationary account for the Mādhyamika, according to Kamalaśīla, is that it is no longer clear in what sense it provides an account of truth. Because a deflationalist tries to account for truth by deflating the world-statement relation, the importance of ‘reality’ gets lost. What is left for an account of truth might then be just beliefs and opinions that people happen to have. As Tille-mans suggests, Candrakīrti’s lokaprasiddha account faces exactly this problem (Cowherds 2011, ch. 9).
The Mādhyamika might respond by proposing a fictionalist account. If there are no ultimate truths, all we can do is to adopt a make-believe attitude toward truth without assuming that fictional ontology does heavy metaphysical lifting. One way to cash out such a fictionalist account is to appeal to Carnap’s distinction between internal and external questions.12 We can think of a belief as a consequence of setting a framework which contains a mechanism to regulate what is admissible within the framework. It is only within a framework that beliefs can be meaningfully talked about. To ask whether reference relations can account for truth, however, is to step outside of the framework. So no such question can be meaningfully raised. This is not to deny the existence of reference; instead, this shows that such a Carnapian account can explain the working of deflation in a fictionalist account.
If we go for a Carnapian account, however, we are already moving away from the reference-based semantics to a pragmatic account of truth. For Carnap, the question about truth comes down to our practice. It has to do with the “planning and optimization of the future of the species” (Carus 2004, 349). The Svātantrika-Mādhyamikas such as Śāntarakṣita and Kamalaśīla may be adopting the pragmatic account of truth espoused by Dharmakīrti (see Cowherds 2011, 145–146).13 As Finnigan and Tanaka suggest, by adopting a pragmatic account of truth, the Mādhyamika can accommodate norms embodied in Nāgārjuna’s statements in terms of practical efficacy. There may not be a theoretical explanation of the norm’s effectiveness. Unlike a reference-based semantics, a pragmatic account doesn’t specify the ontological basis for determining the truth-values of statements. The Mādhyamika can, nevertheless, point at the practice where the effectiveness is observable. By adopting a pragmatic account of truth, the Mādhyamika can respond to the charge of ineffectiveness in the thesis of emptiness raised by Nāgārjuna’s opponent (Cowherds 2011, ch. 11).
Despite the fact that there are several options that Mādhyamikas can take as their semantic account, it is not clear that they have a settled position about semantics. How to make sense of Nāgārjuna’s claim of having no thesis is a live question. The debates surrounding the no-thesis position were transmitted to Tibet. Two of the most influential philosophers in Tibet, Tsongkhapa and Gorampa, seem to have engaged with this very issue. As we can see from Garfield and Priest’s (2003) identification of paradoxes within Nāgārjuna’s system, how best to understand Nāgārjuna’s thesis of emptiness is still debatable. Whatever we can say about Nāgārjuna’s semantics embedded in his thesis of emptiness, one thing is clear: we can see the difficulty of making sense of emptiness once we consider the semantics that must underlie it. We can all agree on Nāgārjuna’s ontology. Even then, we can disagree about his semantics. The realization of emptiness thus depends on the attainment of a semantics that can accommodate it.
1 An English translation of the text can be found in Garfield (1995) from the Tibetan translation, Kalupahana (1986), and Siderits and Katsura (2013).
2 I adopt the modern convention of using ‘Madhyamaka’ for the thought or school and ‘Mādhyamika’ for the thinker.
3 This text is usually not considered to be canonical. However, it is a useful text in explaining Buddhist metaphysics and modern scholars often refer to it.
4 An English translation of Chapter 9 of Abhidharmakośa where Vasubandhu pre-sents an argument for this ontology can be found in Duerlinger (2009).
5 See a discussion by Mark Siderits in Cowherds (2011, ch. 10).
6 See a discussion by Guy Newland and Tom Tillemans in Cowherds (2011, ch. 1).
7 I mainly talk about statements here as truth-bearers. Nothing I say here depends on this, however.
8 See Garfield (1996) and Westerhoff (2009, ch. 9). Garfield claims that this orthodox account is Fregean. Given that it does not posit anything intermediary between the statement (or the thought expressed by the statement) and the referents, that is, sense, the orthodox semantics cannot be Fregean.
9 For the options discussed in this section, see a discussion by Graham Priest, Mark Siderits and Tom Tillemans in Cowherds (2011, ch. 8).
10 See also a discussion by Tom Tillemans in Cowherds (2011, ch. 9) for this interpretation of Candrakīrti.
11 An English translation of this passage can be found in Cowherds (2011, 153–154).
12 For an application of Carnap’s distinction to the two truths, see a discussion by Bronwyn Finnigan and Koji Tanaka in Cowherds (2011, ch. 11). For other fictionalist accounts in the context of Candrakīrti’s lokaprasiddha account, see a discussion by Tom Tillemans in Cowherds (2011, ch. 9).
13 Whether or not Dharmakīrti was a pragmatist is a hard question to answer. For a discussion, see Dreyfus (1997, ch. 17).