Causation and Emptiness in Early Madhyamika

CAUSATION AND EMPTINESS IN EARLY MADHYAMAKA
Author(s): MARK SIDERITS

Source: Journal of Indian Philosophy , August 2004, Vol. 32, No. 4 (August 2004), pp.
393-419
Published by: Springer
Stable URL: https://www.jstor.org/stable/23496964
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide
range of content in a trusted digital archive. We use information technology and tools to increase productivity and
facilitate new forms of scholarship. For more information about JSTOR, please contact support@jstor.org.
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at
https://about.jstor.org/terms
Springer is collaborating with JSTOR to digitize, preserve and extend access to Journal of
Indian Philosophy
This content downloaded from

81.38.69.71 on Thu, 31 Mar 2022 09:46:50 UTC
All use subject to https://about.jstor.org/terms
MARK SIDERITS
CAUSATION AND EMPTINESS IN EARLY MADHYAMAKA
It has long been recognized that there are important c

connections between causation and emptiness in the th
Nâgàrjuna and his commentators. Not long ago it was wid
that these connections went by way of the idea that the d
origination (prañtya samutpannatva) of all things would
'everything is connected to everything else': that the mu
penetration of cause-effect relations meant that the natu
thing is somehow implicated in the natures of all others. T
the matter is less popular today.1 It is seen to be incompa
the central Madhyamaka claim that the doctrine of empti
to be construed as itself giving an account of the ultimat
reality. But no scholarly consensus has taken its place. Inst
faced now with a variety of interpretations of the connect
of which are decidedly skeptical about the overall cog
Nàgârjuna's project.2 It will be my contention that this si
come about in part because of a lack of clarity concernin
notion of svabhâva and its roots in Abhidharma. By rem
selves of the origins of this notion, we may reach a bett
standing of what it means to say that something is empty, i.
of svabhâva.3 And this in turn may help us better unders
early Màdhyamikas see causation and emptiness as intimat
I shall proceed as follows. First I shall recount the orig
term svabhâva and attempt to explain its relevance for th
maka project. Then I shall turn to the question whet
Màdhyamikas have an argument for their claim that depe
causes and conditions entails lack of svabhâva. An examination of the
evidence in favor of a negative answer will bring out the necessity of
examining Nàgârjuna's investigation of causation in MMK 1. Out of
this examination will emerge my central claim: the conceptual con
nection between causal dependence and emptiness comes by way of
the thesis that the causal relation itself is conceptually constructed.
Once we grant this thesis, then it can be argued that anything that
originates in dependence on causes cannot be ultimately real but must
instead be empty of svabhâva.
Journal of Indian Philosophy 32: 393-419, 2004.

© 2004 Kluwer Academic Publishers. Printed in the Netherlands.

81.38.69.71 on Thu, 31 Mar 2022 09:46:50 UTC
394 MARK SIDERITS
A central Abhidharma project was to make philosophical sense of the

early Buddhist teaching of nonself. The Nikayas are completely clear
on the point that there is no self, but they are less clear about the status
of the person - the psychophysical complex that is thought by many to
be the entity that has a self as its essence. The distinction between the
two truths, conventional and ultimate, helps clarify the status of the
person. Since it is a whole made of parts, the person may be said to be
conventionally real but must be considered ultimately unreal. This is
because all wholes are said by orthodox Abhidharmikas to be mere
conceptual fictions (things having only prajñaptisat). These are entities
that are thought to exist only because our discourse employs certain
terms as convenient designators - words that are used to designate a
collection of things related in a certain way . The stock illustration of
this is the chariot. It is conventionally true that there are chariots, and
that King Milinda rode to his meeting with Nàgasena in one. But the
statement about Milinda cannot be ultimately true, since words like
'chariot' (and more importantly, 'Milinda') are mere convenient des
ignators, terms that have been introduced into our (conventional)
discourse as a way of designating a set of entities (e.g., the wheels, axle,
etc.) when these are arranged in a certain way {viz. the 'assembled
chariot' way). While these more particular entities might be thought to
be ultimately real, the whole that they constitute when so arranged
cannot be. For the chariot is an aggregate, something compounded
out of parts, and aggregation or compounding is the mark of the
mental. It is the mind and not reality that dictates that we see these
parts when so arranged as one thing. Otherwise how are we to explain
the fact that we do not see the same set of parts as one thing when they
are arranged somewhat differently - the axle used as a pillar to support
the roof of the shed, one wheel used to form a gate in the fence, the
other wheel used in a sculpture to indicate the Wheel of the Law? Our
use of the word 'chariot' is clearly determined by the fact that we have
a use for the parts when arranged in a particular way, plus the fact that
our minds have a limited ability to track parts in complex aggregates.
But these are facts about us, not facts about the reality of the parts. So
the chariot, and more significantly, the person, are mere conceptual
fictions; only the parts and their relations to one another may be said
to be ultimately real.
Of course, on reflection it is apparent that the wheel is itself a
whole made of parts, so it likewise cannot be said to be ultimately

81.38.69.71 on Thu, 31 Mar 2022 09:46:50 UTC
CAUSATION AND EMPTINESS 395
real. This led the Àbhidharmikas to search for the genuinely

entities, the dharmas, that may truly be said to be ultimately
culminates in a variety of lists provided by different scho
dharmas of the Sautrântikas, the 89 dharmas of the Therav
But behind the disagreements reflected in these differing nu
a key idea that all agree on: only that is a dharma that has sv
intrinsic nature. It is the svabhàva criterion of dharma-hood that
unifies the projects of the Abhidharma schools.
While one still sees the term svabhàva translated as 'own-being' or
'self-existence', it is increasingly recognized that it should instead be
rendered as 'intrinsic nature' or 'intrinsic essence'. The mistake may
have been the result of considering only the Madhyamaka uses of the
term and not tracing these back to their origin in Abhidharma. There
the basic idea was that the properties of something that is ultimately
real cannot be 'borrowed' from other things, but must be intrinsic to
the thing itself.4 The chariot again provides a useful illustration. Its
properties, such as its capacity to transport driver and passengers,
may be accounted for entirely in terms of the properties of its parts.
There is nothing we can say about the chariot that cannot also be said
(albeit at much greater length) without mentioning the chariot and
only mentioning its parts and their relations.5 Of course the same
may be said of the wheel. But if there is such a thing as a reality whose
nature is independent of superimposed concepts that reflect our
interests and limitations, then this regress must end somewhere. And
its terminus will be in things whose natures are genuinely their own
and not borrowed from their parts. These will be the dharmas, and
only they will be ultimately real.
We can now see how the mistranslation of svabhàva as 'self-exis
tence' might have arisen. Mâdhyamikas often claim that something
that is subject to origination and cessation must lack svabhàva and so
must be empty. (Nàgàrjuna so argues in MMK 13 and MMK 15.)
Since it is not always clear that there is any argument behind this
claim, it might be taken as having appeared simply self-evident to
Mâdhyamikas. And this would be the case if by svabhàva they had
understood 'own-being' or 'self-existence'. For then to attribute
svabhàva to something would mean that its existence was due to that
very thing itself. And this in turn at least strongly suggests that the
thing cannot have originated, so that its existence seems to be nec
essary.
Now Mâdhyamikas do hold that anything subject to dependent
origination lacks svabhàva. But this is not because by that term they

81.38.69.71 on Thu, 31 Mar 2022 09:46:50 UTC
396 MARK SIDERITS
meant anything like 'self-existence

the Âbhidharmikas meant: intrin
also held that the dharmas (whic
svabhâva) are subject to dependent
owe us an argument to establish t
originated must be devoid of svab
argument, which we shall come to
say a word in defense of the notio
Abhidharma and Madhyamaka - t
ultimately real it must have svabh
there are those who accuse Mádhy
nents with a highly implausible s
term, so that their arguments tu
fallacy. Hayes (1994: 311), for instance, seems to think that
Nàgàrjuna has set things up so that the 'intuitively correct' view that
complex things originate from collocations of simple things is simply
unavailable to the opponent. But if the svabhâva criterion is justified,
then a complex thing is not ultimately real. So whatever we might
establish concerning its origination could not tell us anything about
how the world is independently of the conceptual shortcuts we use to
attain our goals. If we want to know about origination in the real
world, then we shall need to ask whether simple things - dharmas,
things with svabhâva - can originate from other simple things.
Is the svabhâva criterion justified? I think this criterion should
recommend itself to anyone who is a realist. By 'realism' I shall here
mean just the view that there is such a thing as how things are
independently of the concepts we happen to employ. Realists thus
believe in a reality that is mind-independent in the sense that its
nature is not conceptually constructed.6 Note that on this usage,
Vasubandhu and Berkeley both count as realists. Both Vasubandhu
(in his Yogàcâra phase) and Berkeley deny that things exist inde
pendently of their being cognized. So on their view existing things
have a kind of dependence on minds or the mental. But this sort of
mind-dependence is compatible with the sort of mind-independence
that realists champion. The question whether things may exist unc
ognized is clearly distinct from the question whether things have a
nature that is independent of those concepts that happen to have
proven useful to creatures like us. Subjective idealists like Vas
ubandhu and Berkeley agree with such dualists as Descartes and the
Sautrântikas, and such physicalists as David Lewis and the
Cârvâkans, on their answer to the second question. What they dis

81.38.69.71 on Thu, 31 Mar 2022 09:46:50 UTC
agree with dualists and physicalists about is just whether thing

exist apart from our consciousness of them, not whether th
natures apart from how we happen to conceptualize them. W
calling realism is the view that things do have such natures
can now see, it is commitment to realism in this sense th
pressed in the Abhidharma idea of the ultimate truth. To s
the ultimate truth one must assume that there is such a thi
the world is independently of our conceptual activity.
Now in our previous discussion of the case of the chario
why partite things like it will not be discussed at the level of
truth. But we can now go further and show that ultimately re
must also be simple in their natures. If the ultimate truth is t
of all concepts reflective of our interests and limitations, it ob
must abjure the use of such clearly aggregative concepts a
'bundle', 'forest', and the like. But suppose we think of wa
simple substance whose nature is to be wet or liquid.7 And
we also believe that an individual quantity of this substance
one time be hot and at another time be cold. In doing so w
implicitly introduced the distinction between the essential p
and the incidental properties of a substance. For we are th
posing that the water in question must continue to be liquid
to continue to exist, but that it may be hot at one time an
another. We have thus introduced complexity into our account
seemingly simple substance. Not indeed the same sort of co
that we find in the case of the chariot, where we have one
made up of the distinct substances that are its proper parts
complexity is one of natures, not of entities. Still this me
water as thus conceived must likewise be conceptually con
and not ultimately real. For not only are we then attr
multiplicity of properties to the substance. We are also int
the problematic concept of capacity or potentiality: the
something (viz. something wet) that is potentially hot or p
cold. This smacks entirely too much of the mind's handiwo
the experience in one place first of liquidity conjoined with
then of liquidity conjoined with heat, for the sake of con
economy we introduce an unexperienced 'something' that
and unifies the distinct experiences through the relations of
and incidental inherence. The resort to conceptual econom
same here as in the case of the mental construction of the cha
realist will be justifiably suspicious of this account of water. If
to speak of an ultimately real element in this case, we wou

81.38.69.71 on Thu, 31 Mar 2022 09:46:50 UTC
398 MARK SIDERITS
to think of it as just the occurrence of liquidity. Then wha

ventionally think of as hot water will consist of a mixture
(liquidity element) and fire (heat element); cold water will b
liquidity without the fire element.
The upshot is that for the realist the properties of an ultima
entity must be intrinsic to that entity. Any predicate that
reference (explicit or implicit) to other things, such as bei
atta's favorite color, or occurring on a certain date, will tur
express not a property at all but a relation.8 Since heat is
thought of as intrinsic to fire element, 'hot' cannot be pred
water element; what we call hot water must instead involve
of contiguity between water element and fire element, each wi
intrinsic natures of liquidity and heat respectively. This st
taken together with the prohibition on aggregating properti
that ultimately real entities are such as can only be kn
acquaintance and not by description. This is the point th
Tillemans (2001: 7-9) to express when he parses svabhâva as
that is 'findable under analysis', and contrasts this with
explication of the term using the more general notion of ident
if we understand analysis as breaking something down
constituents of which it is composed, and we see compositio
mark of the mental, then the ultimately real will turn out t
which allows of no further conceptual analysis. Thus the ul
real will turn out to be just those entities whose natures are ex
in whatever monadic predicates apply to them. While ever
whether simple or complex, has identity, only the identities of
remain when analysis has been fully carried out. A co
description of reality will then consist of a list of all such enti
their intrinsic natures, together with an account of the rela
which each of them enters. This is what the Abhidharma schools
sought to give in their theories of the dharmas 9 But the basic idea is
one that I think every realist ought to embrace. Mind-independent
reality consists of just those things whose full natures are independent
of the natures of other things.
ii
The lesson to be learned from the case of the chariot may be put as
follows: anything whose nature can be accounted for wholly in terms
of the properties of the parts of which it is composed is not ultimately

81.38.69.71 on Thu, 31 Mar 2022 09:46:50 UTC
real - it will disappear upon complete analysis. To put the

current terminology, since the property of being a char
supervenes on the properties of its proper parts, the pro
a chariot is reducible to those base properties, and sh
eliminated from our ultimate ontology.10 Now the m
tution relation that obtains between chariot and par
termed a kind of causal relation. This is the sort of relation that
Aristotle dubbed 'material causation', the Naiyàyikas theorized as
samavâyikàrana, and Samkhya referred to as upadanakarana.u We
may then say that for this type of causation, anything that is causally
dependent on other things lacks svabhâva and so is not ultimately
real. This much is common to Abhidharma and Madhyamaka. The
question is whether this holds as well for other sorts of causal relation
besides that of material constitution, whether it holds for the causal
relation in general. This is where Madhyamaka parts company with
Abhidharma. And this is where Hayes sees a major logical flaw in
Nàgàrjuna's argument. In effect, Hayes accuses Nàgàrjuna of having
overlooked the difference between material causation and other
forms of the causal relation in order to arrive at the conclusion that
causal dependence entails lack of intrinsic nature. Tillemans (2001:
16-24) responds to this charge by citing arguments from later
Madhyamaka sources that appear to establish a conceptual connec
tion between causal dependence in any sense and lack of intrinsic
nature. But Hayes might reply that there are passages in MMK
suggesting that Nàgàrjuna thought going from one to the other re
quired no argument-thus bolstering Haye's charge of equivocation
on Nàgàrjuna's part.
Hayes relies largely on MMK 15 to substantiate his charge of
equivocation. And we can see why, given that in 15.2ab we have
Nàgàrjuna saying, 'And how could there ever be an intrinsic nature
that is a product?' (svabhâvah krtako nama bhavisyati punah katham)
without appearing to see the need for an argument. Still Bhàvaviveka
and Candraklrti do provide arguments for the claim that an intrinsic
nature could not be a product.12 (The earlier commentaries Akutob
hayà and Buddhapalitavrtti merely repeat the claim in slightly dif
ferent words; see Pandeya 257.) It is thus difficult to say whether
Nàgàrjuna actually had such an argument in mind. What is perhaps a
clearer case of the apparent problem occurs at MMK 4, where
Nàgàrjuna uses the example of rüpa skandha to argue for the con
clusion that all five skandhas are empty.13 The argument for the
emptiness of rüpa depends on the fact that rüpa or corporeality (those

81.38.69.71 on Thu, 31 Mar 2022 09:46:50 UTC
400 MARK SIDERITS
ultimate entities grasped by the five external senses) is

Àbhidharmikas to be causally dependent on the mahñbhütas
material elements of earth, water, fire and air. Now in th
Nàgàrjuna does provide an argument for the claim that
dependence entails lack of intrinsic nature (MMK 4.2-6). Bu
read this argument without the benefit of the later comme
Bhàvaviveka and Candraklrti, then whatever plausibility it
pears to derive from the fact that the relation between rüp
mahñbhütas seems to be one of reductive supervenience. N
argues that rüpa could not be seen as distinct from its cause
that case it should be possible for it to occur independentl
occurrence of the four elements, which is incompatible with th
that rüpa has the four elements as its material-base cau
isthâhetu).14 But if rüpa were identical with the four elements
should turn out to be a mere aggregate of the four elements, in
case it would not be ultimately real and so must lack intri
ure.15 Now it might be thought that the Àbhidharmika ca
retreat to the view that rüpa really just is the four element
case, while rüpa could no longer be said to have the four ele
cause (it being the four elements), rüpa could still be said
intrinsic nature and so be ultimately real.16 But this will no
the four elements only occur as atoms and are thus imperc
is crucial to the realism of Abhidharma that the ultimatel
perceptible; to allow mere theoretical constructs (such as i
tible atoms) into our ultimate ontology is to allow the min
much power in determining what the world is like.17
So Nàgàrjuna's argument appears to succeed in showing t
lacks intrinsic nature. The difficulty is that in MMK 4.7, h
generalize the argument to the case of the four non-materi
skandhas. Now each of these may also be said to arise, like
dependence on causes and conditions. But the causal rela
importantly different in their case. Consciousness, for instance
to arise in dependence on sense-object contact. But no Àbhid
would want to say that sense-object contact is the material-b
of consciousness. This is because for five of the six typ
sciousness, sense-object contact is thoroughly physical i
sense faculty and sense object both belong to rüpa skan
contact is a relation between the two. And to say that consc
reductively supervenes on something wholly physical in nat
embrace physicalism and deny a distinct place for the ment
ontology. The causal dependence that consciousness has

81.38.69.71 on Thu, 31 Mar 2022 09:46:50 UTC
object contact is of a completely different kind than the material

constitution dependence of rüpa on the four elements. So when
Nàgàrjuna tries to generalize from the case of rüpa to the other
skandhas, he seems to be committing just the mistake that Hayes
accuses him of.
in
Does Nàgàrjuna have an argument for the claim that causal depen
dence in general entails lack of intrinsic nature? I think he does. But
to find it we need to first go back to the discussion of causation in
MMK 1, and then think through the consequences of its conclu
sion.18 This chapter begins with the famous claim that an existent
cannot be said to originate from itself, from what is other, from what
is both itself and other, and from no cause at all. It is generally agreed
that the first and second lemmas represent the alternatives of
satkñryavñda and asatkaryavâda respectively that were widely de
bated among Indian philosophers concerned with the nature of the
causal relation. According to the former view, the effect pre-exists in
unmanifest form in its (material) cause. According to the latter, cause
and effect are distinct entities with distinct natures. Now Hayes (1994:
313) concedes that the prospects of the first alternative are dim. It is
hard to make out what it would mean to say that the effect already
exists prior to its production; to be told that it only exists as a
potentiality is not really to be told any more than just that the cause
will bring it into existence, something we presumably already know.
What Hayes cannot see is why Nàgàrjuna thinks the second alter
native must be rejected. He takes the argument that is supplied in
MMK 1.3cd, 'The intrinsic nature not occurring, neither is the
extrinsic nature (parabhâva) found', to be little more than a play on
words that exploits the ambiguity of the Sanskrit bhâva: 'Something
whose being (bhava) is dependent on another cannot have a nature
(bhâvà) that is its own'. This leads him to see the argument of the
three times (MMK 1.5-7) as a distinct argument against the causal
relation. But this misses the thread of the argument against
asatkaryavâda.
Let us look more closely at how Nàgàrjuna actually argues. The
relevant verses are these:
1. Not from itself, not from another, nor from both, nor with
out cause, Never in any way do there obtain any existing
things that have arisen.

81.38.69.71 on Thu, 31 Mar 2022 09:46:50 UTC
402 MARK SIDERITS
2. There are four conditions: the primary cause, the o

support, and the proximate condition, And of cour
dominant condition; there is no fifth condition.
3. The intrinsic nature (svabhava) of existents is not foun
conditions etc. The intrinsic nature not occurring, nei
the extrinsic nature (parabhâva) found.
4. An action (kriya) does not possess conditions, nor is it
of conditions. Conditions are not devoid of an action,
are they provided with an action.
5. When something arises dependent on them, they are
be conditions. When something has not originated, wh
are they not non-conditions?
6. Something cannot be called a condition whether th
[which is supposedly the effect] is [already] existent
[yet] existent. If non-existent, what is it the condition of
if existent, what is the point of the condition?
7. Since a dharma does not operate when [the effect] is e
non-existent, both existent and non-existent, how in
case can there be an operative cause?
In verse 1 Nâgàrjuna lays out the overall strategy of the arg

will give in this chapter. (He does this elsewhere in MMK, e
outset of Chapters 8, 10, 12, and 16.) In the remainder of th
he will present arguments against the first and second lem
elsewhere, so here too he will not explicitly argue against
and fourth lemmas, whose absurdity should be obvious onc
the difficulties with the first and second. In verse 2 he then gi
standard Abhidharma classification of types of causal facto
will be the target view as to how ultimately real entities could
origination. In 3-7 he will argue against the notion of the
cause (hetu-pratyaya), while in the remaining verses of the c
will give arguments against the other three types of conditi
As Candrakïrti makes clear, in demi-verse 3ab Nàgàrjuna
senting an argument against satkàryavada, the theory that t
exists in unmanifest form in the cause.19 The verse states
intrinsic nature of the effect is not to be found among the con
that serve as its cause. This is tantamount to a rejection of
satkàryavada because to say that the effect exists somewhere among
its conditions is to say that something with the intrinsic nature of the
effect, for instance the heat of fire, is to be found among the condi
tions of fuel, air, absence of moisture, striking motion, etc. Not only

81.38.69.71 on Thu, 31 Mar 2022 09:46:50 UTC
is this not true, it is clearly absurd. Typically we want to k

cause of something because we want to know how to produc
we desire it. If the nature of fire were already present in
another of its conditions, or in all collectively, we would not b
try to produce fire. Satkâryavâda might do as an accoun
material constitution relation between the cloth and its threads. But
precisely because the cloth is thereby shown to be an aggregate, this is
not a relation among ultimately real entities, which is what we are
after.
The demi-verse 3cd has proven harder to decipher. While it liter
ally reads as a denial of extrinsic nature in general, the context (an
examination of causation) makes this reading implausible, so it is
generally agreed that it is intended to deny that the effect originates
from conditions that are in some sense extrinsic to it. This suggests
that the target view is asatkaryavada, the view that cause and effect
are distinct entities. But then there is the question what the argument
is against this view. The line may be read as claiming that the disproof
of arising from distinct things is a consequence of the refutation of
arising from identical things. Several modern commentators follow
this approach. Taber (1998: 216f), for instance, says the argument
depends on what he calls 'the principle of coexisting counterparts', to
the effect that an entity whose nature is in essential relation to some
other thing cannot exist apart from that thing. The argument would
then be that since the notion of 'distinct' requires the notion of
'identical', and it has already been shown that there can be no arising
from what is identical, it follows that there can be no arising from
what is distinct.20 But this is a singularly bad argument. While
Nàgàrjuna might be seen as employing something like the principle in
question elsewhere, it is of dubious legitimacy. And in any event this
argument involves its misapplication. It is the properties of being
identical and being distinct that may be said to be counterparts, not
the properties of originating from what is identical and originating
from what is distinct. A Mâdhyamika might legitimately argue that
if there is no intrinsic nature (svabhàva), there can be no parabhava
or 'borrowed' nature either, since there would then be no natures
to be borrowed. But Nàgàrjuna has not yet given evidence for
the non-existence of intrinsic nature. All he has argued for is that
entities cannot originate from something with their own intrinsic
nature. So if there is any way of avoiding attributing the argument
Taber sees here, the principle of charity in interpretation dictates that
we do so.

81.38.69.71 on Thu, 31 Mar 2022 09:46:50 UTC
404 MARK SIDERITS
The classical commentators suggest another way of con

argument. Buddhapàlita takes the demi-verse to simply
origination cannot be from what is other, and sees the ar
this claim as coming in the succeeding verse 4 (Pande
fails, however, to account for the syntactic construction of
suggests a connection between the rejection of originatio
and rejection of origination from other. CandrakTrti, wh
that 4 supports 3cd, helps us see how 3cd might be c
3ab.21 The strength of satkâryavâda lies in the fact that
to see why an entity with a certain nature should have
rejection in favor of origination from conditions constru
or extrinsic means that we lack an account of the neces
tion between cause and effect. We know that consciousness does not
arise from cloth. Why then should it arise from sense-object contact,
since, as CandrakTrti points out, its nature is not found there? Of
course we know that sense-object contact has regularly been followed
by the occurrence of consciousness, and that the mere existence of
cloth has not been regularly followed by consciousness. But to say
that consciousness is caused by sense-object contact (and not by the
existence of a piece of cloth) is to say that the regularity we have
observed is a robust trend that may be counted on to continue into
the future, and is no mere fortuitous string of coincidences. So some
explanation is called for as to why we should expect the one thing to
always follow the other. If satkâryavâda were true we would have an
explanation ready to hand: in the one case the nature of the effect is
already there among the conditions, so it is no mere accident that the
effect always arises from these but not from the others. But the
asatkâryavâdin can have no such recourse, and so still owes us an
explanation. This is the point Nâgârjuna seems to be getting at when
he says that since the intrinsic nature of the effect is not in the con
ditions, it will not do to say that the effect arises from something with
a distinct nature (that the cause is parabhâva to the effect).
All this so far has just been a retracing of the usual dialectic of
causation in Indian philosophical discussions. What comes next is
new and interesting. The Abhidharmika will naturally be an
asatkâryavâdin, like the Naiyàyika and unlike the Sâmkhyan and the
Advaita Vedântin. The alternative of satkâryavâda is ruled out for the
Buddhist since it clearly leads to eternalism. We have seen that the
difficulty for the asatkâryavâdin is to explain why the effect should
arise from just this array of conditions given that none of them is of
the nature of the effect. The obvious move for the asatkâryavâdin to

81.38.69.71 on Thu, 31 Mar 2022 09:46:50 UTC
make at this point is to invoke a causal relation to do the

explanatory work. To say that consciousness arises from se
contact is then just to say that when there is contact betwe
object and sense of vision (under suitable circumstanc
assembly of conditions yields a causal force that in turn gives
moment of consciousness.22 This is then said to be why the reg
we have observed is no mere coincidence but a robust trend: we have
been observing a causal law at work, a causal necessity or power built
into the nature of the world such that these conditions must produce
this effect. In the Nyàya formulation of asatkâryavâda it is samavàya
that is usually invoked to do this work, but here Nàgarjuna has the
Àbhidharmika call this causal connection a kriya or action.23
In verse 4 we find an argument against this idea of a causal power.
Candraklrti's and Buddhapâlita's commentaries have it that the
argument is just an instance of the 'three times' argument against
causal production that is presented in 5-7, and to which we shall
come shortly. But I wish to explore another possible interpretation
that is suggested by the verse.24 Recall that the verse claims an action
can neither possess nor fail to possess conditions, and likewise that
conditions can neither possess nor fail to possess an action. The
asatkaryavâdin opponent began with two things - the collection of
conditions and the effect - and wants to explain why the first
invariably gives rise to the second. The proffered explanation is that a
causal force mediates between them. But now we have a third thing
appearing on the scene, and it is legitimate to ask how it comes to be:
is it likewise produced in dependence on conditions, or not? If not,
then it is uncaused and so presumably unreal. If it is, then we may ask
further whether it already exists among those conditions or not. If it
does, then we have reverted to the discredited theory of satkâryavâda.
If it does not, then the logic of the situation requires that we posit yet
another causal force, causal force2 to mediate between conditions and
the original causal force i in order to explain why just this sort of
thing (a causal force capable of producing just the effect of con
sciousness) should arise given the occurrence of these conditions. And
now an infinite regress yawns before us: if causal force2 is required in
order to explain why causal force j should arise from its conditions,
then we shall need yet another causal force, causal force3, in order to
explain why just the right sort of causal force for producing causal
forcej should arise from these conditions.25
This is no mere piece of sophistry, the difficulty here is quite real. If
cause and effect are truly distinct and of separate natures, we need

81.38.69.71 on Thu, 31 Mar 2022 09:46:50 UTC
406 MARK SIDERITS
some account of why we are justified in supposing that gi

cause, the effect will occur.26 Suppose we have frequently seen
sort of event followed by the other. We must still ask wh
justified in projecting from the past association of the two e
future occurrence. To seek to justify this projection on the bas
assumption that the future will be like the past is, as Hum
reason in a circle. Perhaps we are inclined to believe that th
will be like the past because in the past this belief was born
that is no argument, since the conclusion follows only g
additional assumption that the future will continue to be like t
So the demand for an explanation of the regularity must be
and the posit of a causal force looks like the only choice. B
posit fails to satisfy the demand, since it leads to a new
questions that looks unlikely to ever end.
Now it is open to the asatkâryavâdin to object that Nàgà
misconstrued the hypothesis. To speak of a causal conn
tween conditions and effect is not to speak of some new th
comes between conditions and effect. A causal connection is not a
thing but a relation, it might be said; and relations do not stand in
need of yet other relations in order to hold between their relata and
thus link them.27 The relation of being next to holds between one
page of the book and another, but we do not need yet another 'next
to' relation to tie the first page to this relation, nor a third to tie the
relation to the second page. If we did, then the two pages would not
be next to one another after all, since there would be infinitely many
next-to relations between them. The relation just simply connects the
two relata without needing any further relations to tie it to its relata.
(The Naiyàyika would say the relation is 'self-linking'.) So let us
grant that the causal relation intrinsically links cause and effect
without giving rise to infinite regress problems. In any event,
Nàgàrjuna has a new question to raise.28
In verses 5-7, Nàgàrjuna gives two somewhat different formula
tions of what can be called the argument from the three times. (He
uses the same pattern of reasoning elsewhere, e.g., in MMK 2 and
MMK 3.) The argument is basically this: if a set of entities and events
is to be considered collectively the /zeiM-condition of some effect (i.e.,
its 'total cause'), there must be some time at which this set produces
the effect.29 But this time cannot be the period when the effect exists;
for then the effect has already been produced, so that production
would be superfluous. (6d: 'And if existent, what is the point of the
condition?') Nor can it be when the effect does not yet exist; the

81.38.69.71 on Thu, 31 Mar 2022 09:46:50 UTC
absence of the effect at that time shows that the conditions have not
yet produced any result. (5cd: 'When something has not originated,
why then are they not non-conditions?' 6c: 'If non-existent, what is it
the condition of?') And there is no third time between the time when
the effect does not yet exist and the time when the effect does exist
(7ab: 'a dharma does not operate when [the effect] is ... both existent
and non-existent'). One can supply a third time, the time when the
effect is undergoing production, only by supposing that the effect is
partite and hence not ultimately real. Here it is crucial to recall that
what we are seeking is the ultimate truth concerning the causal
relation; and this must obtain among things that are themselves
ultimately real. Since ultimately real things are impartite, they cannot
be said to undergo a temporally thick process of production; at any
moment, either such an entity exists or it does not. A chariot may
undergo such a process: first there are just the scattered parts, then
some are put together with others to form significant sub-assemblies,
and when enough of these are put together the chariot comes to be.30
But something simple, a dharma, cannot emerge into existence a bit at
a time; either it's there or it is not.
Some care is called for if one is to appreciate the force of this
argument. It is particularly important to bear in mind that the
satkâryavâdin's challenge to asatkâryavâda is still in force. Given that
cause and effect are distinct entities, what is the necessary connection
between them such that the one results in the other? The causal realist
will want to reply that the cause simply produces the effect, as a
potter produces a pot. The effect has the nature it has because it was
produced by a cause with a certain nature of its own, just as this sort
of pot came into existence because the potter had these desires and
performed these actions. But this model of the production relation
requires that the effect gradually emerge through a production pro
cess that occurs over an extended period of time. We see how the
potter could control the nature of the pot because we picture the
potter shaping the clay on the wheel, applying glaze, and then firing it
overnight. And this model is unavailable to the causal realist in the
case of ultimately real entities. If the production relation is to supply
the missing explanatory link, it must relate cause and effect at some
time other than when the effect is undergoing production. This leaves
the time when the effect exists, and the time before the effect has come
into existence. The first alternative is clearly a non-starter. And the
trouble with the second is that it requires that a relation connect an
existing thing with a non-existent. An existing cause can no more

81.38.69.71 on Thu, 31 Mar 2022 09:46:50 UTC
408 MARK SIDERITS
produce a non-existent effect than a real person can be ador

a garland of sky-flowers, or a real house can be inhabited b
of a barren woman. We suppose the first case is different b
think of the effect as something that will exist in the future.
fail to note is that it is then our expectations that supply th
relatum. That should warn us that the production relati
envisage is conceptually constructed.
There are other objections that might be raised to this ar
One might, for instance, claim that there is after all a third ti
the conditions may be designated as cause of the effect, nam
both exist. That is, why cannot cause and effect be strictl
neous in occurrence?31 If cause and effect arise together, then
not be said that the occurrence of the cause produces the oc
of the effect? As Burton (1999: 193) points out, it was the o
Abhidharma view that this is possible. Thus Vaibhâsika
ravàda hold that each of the four material elements is in r
causal interaction with the others, such that none can exis
pendently of the others. And while Sautrântika denies r
causal interaction, it does allow that cause and effect may b
at the same time. Burton thus faults Nàgàrjuna for having
imposed by stipulation that cause must precede effect. But w
true that Nàgàrjuna requires that cause precede effect, ther
good reason for this. Consider an apparent case of reciproc
interaction: the raising of one cup of the balance and the lo
the other, each of these two events occurring simultaneou
Màdhyamika will then ask which is the cause and which the
say that each is both cause and effect of the other is no b
answer than to say that neither one is cause of the othe
because both are in fact the effects of some third event, the p
a weight in one cup, an event that precedes both. And in ca
as that of the interweaving of the threads and the appearan
cloth, where there is simultaneous occurrence without app
ciprocal interaction, but where one might be seen as cau
other, we have material constitution again, and so not a
between ultimately real entities. It is not clear that there c
causation between events that are simultaneous in occurrence.
Suppose, then, we grant that the set of conditions precedes the
effect. (This must be true if by 'set of conditions' we mean 'total
cause': among the conditions for the origination of an entity is the
absence of that entity at a certain time and place; this set ceases to
obtain when the effect comes into existence.) Thus when the set of

81.38.69.71 on Thu, 31 Mar 2022 09:46:50 UTC
conditions exists, the effect does not yet exist, and when
does exist the set of conditions no longer exists. But why s
this rule out our saying that the conditions produce the effec
is the error in seeing the causal relation as something joi
distinct things at distinct (though contiguous) times? In fa
no error in this-provided we are willing to concede that
relation is conceptually constructed. This was Hume's c
and Bhâvaviveka concurs: 'However, parabhava of the con
found due to intentness of the mind on the desire for w
ductive of the arising of bhava - i.e., just by virtue of ex
(Prajnàpradïpa on MMK 1.3, Pandeya 26). That is, our exp
the past association of a set of conditions and an effect l
expect something with the nature of the effect to arise ou
(distinct) conditions, and the causal relation is just our pro
that expectation. Only the intervention of the mind can ex
two things occurring at distinct times can come into so
relation. The only alternative is to embrace a block un
ception of time, according to which all three times exist t
some timeless sense of 'exists'. And as the Sarvàstivàdins d
this renders impossible a coherent account of change, and t
end leads to eternalism.
Taber would no doubt disagree. He agrees that the satkârya /
asatkaryvada problematic constitutes the framework for Nâgàrjuna's
arguments. But he complains that Nàgàrjuna simply uses the objec
tions of adherents of one theory to attack the other, without con
sidering the 'most reasonable' position that certain cases of causation
might be amenable to treatment under one model, while others are
better accounted for in terms of the other, or exploring the possibility
that there might be refinements of either theory that would enable it
to circumvent the objections raised against it (1998: 221). But while
open-mindedness and flexibility are no doubt intellectual virtues, it is
not clear that they will help us evade Nâgàrjuna's conclusion. For
once we accept the claim that partite entities are conceptually con
structed, then satkaryavada is ruled out as an account of causal
relations among ultimately real things. Yet the satkaryavadin has a
legitimate complaint against asatkñryavñda: if cause and effect are
truly distinct ultimate reals, we need some account of the necessary
connection between them.
That this demand is not easily fulfilled may be seen by looking at
current attempts to respond to it. The most popular approach seeks
to use model-theoretic semantics to explain what sort of necessity is

81.38.69.71 on Thu, 31 Mar 2022 09:46:50 UTC
410 MARK SIDERITS
involved here. It is clearly not logical necessity: Hum

right that the existence of the effect is not logically en
occurrence of the cause. Nor does it seem to be metaphy
sity, or invariable concomitance in all possible worlds. F
imagine possible worlds in which the causal laws are qui
from those obtaining in the actual world. Thus nomologi
is taken to consist in invariable concomitance of the two
in all relevantly similar possible worlds. The difficulty w
swer is that it does not seem particularly informative. F
vantly similar' worlds turn out to be just those worlds
laws like those governing the event-types in question in
world. So even if we knew just what a possible world is,3
not seem much closer to an answer to our question.
The model-theoretic approach is one that will appeal t
nalists. For those willing to grant the existence of univer
the alternative of explaining nomological necessity in t
tions between universals.33 This approach has the advan
needing to connect existent entities with things that do not
of yet, exist. Since universals are eternal entities, the c
tion will be between things both of which exist at any given
this also creates a difficulty: how do eternal universals
connected to their non-eternal instances? In MMK 5,
argues that there is no coherent account of the relat
dhatus such as space and their defining characteristics.
argument was not formulated with the case of inheren
universal and particular in mind, it applies to that relatio
defend the claim that nomological necessity is a relat
universals, one must not only reply to the many objections
existence of universals, but address this difficulty as w
clear that a vindication of causal realism is to be found
proach either.
IV
Nàgàrjuna has, then, some interesting and powerful arguments

against a realist construal of the causal relation. Elsewhere I explore
some of their consequences for epistemology and philosophical
semantics.34 But the question with which we began was a meta
physical question: does causal dependence in general entail lack of
intrinsic nature? The reader may have noted that while Hayes takes

81.38.69.71 on Thu, 31 Mar 2022 09:46:50 UTC
MMK 1 as a whole to be an argument for an affirmative answer to

this question, this is not how I have represented it.35 All that I see the
overall argument as showing is that the causal relation is conceptually
constructed. This does not by itself show that the relata of that
relation are also conceptually constructed. And this is what the
Màdhyamika must show in order to forge a logical connection be
tween causal dependence in general and emptiness or lack of intrinsic
nature. Is there, after all, such an argument?
I do not know of any early Madhyamaka text that explicitly for
mulates such an argument. But I think the conceptual connections
are clear enough to allow us to attribute at least implicit recognition
of something like the following thought.
P: If a relational tie is conceptually constructed, then any property

of one of its relata that involves essential reference to that tie
must likewise be conceptually constructed.
The argument would thus be that since the causal connection is

conceptually constructed, then since the intrinsic nature of an effect
would involve essential reference to the causal relation on which it
depends, the intrinsic nature of the effect would be conceptually
constructed. And since the intrinsic nature of an ultimately real entity
is meant to represent the nature of that entity independent of all
conceptual construction, it follows that something that is causally
dependent could not be ultimately real or have intrinsic nature.
Principle P requires some defense, for on its face it appears vul
nerable to a serious objection. Suppose it were true that the mountain
we know as K2 is the mountain most sacred to the Gandhàrvas.
(Perhaps this is true because all the tales of the Gandhàrvas agreed on
this point, just as it is true of London that it is the city in which
Sherlock Holmes resided because all the tales of Sherlock Holmes
agree on that point.) Now the relational tie 'being sacred to' is con
ceptually constructed. Something cannot be held sacred by some
individual or group in the absence of the whole set of interrelated
concepts and practices concerned with what is broadly considered the
religious. Still this would not seem by itself to make K2 something
that is conceptually constructed. For consider the fact that just as
Sherlock Holmes' being a fictional character does not make London a
fictional city, so the fact that the Gandhàrvas are mythical beings
would not make K2 a mythical mountain, even though K2 had the
property of being sacred to the Gandhàrvas. By the same token it
would seem that the fact that K2 is a relatum of the conceptually

81.38.69.71 on Thu, 31 Mar 2022 09:46:50 UTC
412 MARK SIDERITS
constructed relational tie 'being sacred to' should not

ontological status of K2 itself.
But this is why P specifies that the property in quest
essential reference to a conceptually constructed relatio
essential to the property of being K2 that something be
Gandhârvas, or to anyone. So this case is not a counter-e
What then is 'essential reference'? Well, the property
chariot would appear to involve essential reference to th
having been manufactured by someone. For something
a chariot unless it were an artifact, and artifacts are par
that paradigmatically are created through being manuf
persons. So just as the property of being a flying chari
mythical in the case where the entity in question has the
having been manufactured by the Gandhârvas, so the p
being a chariot will be a conceptual construction if the
being manufactured is conceptually constructed (regard
ther or not the manufacturers are themselves real). And
the property and the relation are conceptual constru
property of being a chariot is generally accepted as a pa
of something that is conceptually constructed. And to s
action of manufacturing, one must make implicit reference
of the process and the intentions of the assembler(s). This
instance, when we describe the beehive as having been m
by the bees, we are implicitly attributing fairly rich co
sources to the bee: the ability to formulate plans and
coordinated actions in accordance with those plans, the a
when a goal has been achieved, etc. Those who see such
as anthropomorphizing will be inclined to say instead th
merely hard-wired to behave as if it were manufacturin
behavior, they say, is tropistic. There could not be the
being manufactured by someone in the absence of certain
resources. And for this reason the property of being a chari
a conceptual construction.
Now Nâgârjuna would seem to hold that the property o
intrinsic nature involves essential reference to the causal connection
that holds between it and the conditions in dependence on which it
originated. We can confirm this by seeing what happens when we turn
the relation of causation into a property by filling one of the 'holes' in
this relational tie. Suppose it were said that a certain entity has its
intrinsic nature as a result of having been caused to be of that nature
by the Gandhârvas. Clearly it could only be said to have such a

81.38.69.71 on Thu, 31 Mar 2022 09:46:50 UTC
nature mythically (that is, the statement could be taken as

within the context of the recounting of a myth). Likewise
said to have its intrinsic nature because Sherlock Holmes caused it to
be of that nature, we would have to take this attribution of nature to
be fictional, something that is only true 'in the story'. This tells us that
reference to the relational tie is essential to the having of the property.
And we have already seen that there is good reason to think this
relational tie is conceptually constructed; this is just the point of the
arguments of MMK 1. It then follows that the intrinsic nature of
anything that is caused to have its nature must be conceptually
constructed. And since an ultimately real entity cannot have a nature
that is conceptually constructed, and an ultimately real entity must
have an intrinsic nature, it follows that nothing that originates in
dependence on causes (in the broadest sense) can be ultimately real.
All Buddhists can agree that something made of parts lacks intrinsic
nature and so cannot be ultimately real. The question at issue between
Abhidharmika and Màdhyamika has just been whether something
that is impartite could depend for its nature on a cause that is
understood as distinct from the effect after the manner of
asatkñryavada. In MMK 15.1-2ab, Nàgàrjuna asserts it could no
Our question has been whether there is an argument available
Nàgàrjuna to back up this assertion. Part of the argument, I have
claimed, is provided by the proof in MMK 1 that the causal relation is
conceptually constructed. The rest depends on seeing what it means to
say that something has an intrinsic nature, and why a realist shou
agree that only things with intrinsic natures are ultimately real. Once
this is seen, then Principle P becomes intuitively plausible, and th
provides the necessary connection between the conceptual constru
edness of causation and the emptiness of things that are caused.
NOTES
1 Taber (1998: 236f) is an exception. My thanks to an anonymous refer

journal for bringing this article to my attention.
2 An interesting and influential recent formulation of such skepticism is
in Hayes (1994), which shall be my main target. In seeking to defend
hyamaka against Hayes' criticisms I am following Tillemans (2001). But
somewhat different strategy here, though one that I think should pro
mentary to his.
3 See, e.g., Candraklrti's comment in Prasannapada on MMK 15.2
prakrtih, yeyam sûnyatâ, naihsvàbhavyam (de la Vallée Poussin 264, Pan
4 See, e.g., AKBh 1.18cd. See also AKBh 6.4, where Vasubandhu spe
ultimately real as that which excludes what is other (anyapoha).

81.38.69.71 on Thu, 31 Mar 2022 09:46:50 UTC
414 MARK SIDERITS
5 Strictly speaking this is not quite right.

will have a fully synonymous counterp
ultimately true statement no matter how
claim that a certain person deserves rewar
e.g., VM XVII.162-172.) What can always
a complete explanation of why it would
statement is (conventionally) true - why it
the statement given our interests and cogn
6 It is a different question whether the
captured using some privileged set of con
this question. Abhidharmikas and Naiyâ
Yogâcâra-Sautrântikas and Advaita Vedâ
understand that the view that ultimate r
variety of realism. From this it follows th
the ultimate nature of reality is not equ
conceptualization. The difference here is
Madhyamaka and the Yogâcâra conceptio
The Madhyamaka claim that all things
hidharma background against which it is m
are conceptual constructions. (Nâgârjuna
or vikalpa in this Abhidharma sense, m
conceptual activity and thus not ultima
Burton (1999: 109f) takes this to commit
whatever exists. I take it instead to amo
Siderits (2003: 157-195) for an attempt a
7 Note that this presupposes the classica
that is common to the sciences of ancient
theory of water as the compound H20. Acc
be solid or gaseous.
8 By a property I shall here mean that wh
a relation I shall mean what is expressed b
So the monadic predicate 'is red' expres
predicate 'is redder thanj expresses the r
first is redder than the second, the triadic
expresses the relation three things stand in
of the second and that of the third in its
9 See VM XI.85-93 for an example of thi
the person is a mere conceptual construc
10 The argument for the elimination cla
include the supervening property in our
properties is to count the same thing tw
realist scruple.
11 The three accounts obviously differ
nothing analogous in the Abhidharma ac
because the material constitution relation
since one of its relata is not ultimately r
given in Sâmkhya, Nyâya, and Yogâcâra
2 Candraklrti, for instance, in effect argu
causes and conditions in dependence on wh
have borrowed its heat from those causes
a substance-attribute (dharma-dharmin) m
respond that fire just is its svabhâva of he
and borrowed to be identified here. But

81.38.69.71 on Thu, 31 Mar 2022 09:46:50 UTC
view that dharma and svabhâva are identical, in his extended

1.1. Candraklrti also gives an interesting account in his comm
how the erroneous view that real things have svabhâva mig
otherwise there would not be recognition by means of wh
characteristic must be its very own. And this being established
the Buddha declared the conventional own-form of these in the Abhidharma.'
(tato'nyatrânupalambhâd asâdhâranatvena svam eva laksanam iti krtvà / balajan
aprasiddyaiva ca bhagavata tad evaisñm samvrtam svarüpam abhidharme vyavasthap
tam (de la Vallée Poussin 261, Pandeya 258)).
13 The skandhas are the five categories of existents recognized in early Buddhist
accounts of the constituents of the person: corporeality, feeling, perception, voli
tional forces, and consciousness. The first is generally set apart from the other four
through the contrast between rüpa skandha (corporeality) and the nama skandhas
(consciousness and the rest), thus acknowledging a kind of dichotomy between the
material and the mental.
14 See Vasubandhu's comments at AKBh II.65b
15 See AKBh II.22d for a debate between the Vaibhàsika and the Sautrântika over
this matter. In the end the Vaibhàsika seems to concede that so-called upâdâya rüpa
(which is what is actually grasped by the senses) in fact consists in a complex
arrangement of atoms of the four elements. In that case, phenomenal properties such
as color and odor will turn out to represent nothing more than how certain com
plexes of atoms of the elements happen to appear to creatures with senses like ours;
they will in effect become mere secondary properties. This view of the relation
between rüpa and the four elements was required of the Abhidharmikas in order
to block the claim of the Pudgalavâdins that the person non-reductively supervenes
on the five skandhas. See AKBh IX, pp. 461 ff. for Vasubandhu's account of this
debate.
16 In MMK 1.13 Nâgârjuna gives an argument to show why material constitution
when construed as a relation of identity cannot count as a cause-effect relation
among real entities:
The product consists of the conditions, but the conditions do not consist of
themselves.
How can that which consists of conditions be something which is the product of
what does not consist of itself?
The purported effect, the chariot, is one thing consisting of, let us suppose, 287 parts.
The parts, however, do not have the property of consisting of 287 parts. (They have
instead the property of being 287 in number.) So if the chariot just is the parts, we get
the absurd result that it both does and does not consist of itself.
17 See the commentary on Vimsatikâ 14d, where the Abhidharmika opponent
(apparently a Sautrântika) concedes that the atom is a mere conceptual construction.
18 In this I agree with Garfield (1994) that it was important for Nâgârjuna to begin
MMK with the examination of causation. I disagree, though, with his claim that in
MMK 1, Nâgârjuna offers a positive account of causal relations as 'explanatorily
useful regularities' (222). This claim rests on the assertion that Nâgârjuna distin
guishes between the hetu, which he rejects in MMK 1.1, and the pratyaya, which he
accepts in MMK 1.2. But this assertion leads to a strained reading of MMK 1.4-5, as
well as to the acute problem that he must then make MMK 1.11-13 objections -
when there is no evidence in any of the extant commentaries that this was
Nâgârjuna's intention. Now I happen to agree that the 'positive' account of causa
tion that Garfield provides is the sort that a Mâdhyamika should hold (at the con
ventional level, of course). Indeed Bhâvaviveka could be read as suggesting just such
a view in his comments on MMK 1.3. (See Pandeya 26.) But I see no reason to
believe Nâgârjuna meant to provide such an account anywhere in MMK.

81.38.69.71 on Thu, 31 Mar 2022 09:46:50 UTC
416 MARK SIDERITS
19 Candraklrti begins his comment on

pratyayesu samastesu vyastesu vyastasamas
cidbhâvânâm kâryanâm utpâdât pürvam
Vallée Poussin 78, Pandeya 26). He thus u
not in the conditions to mean that the effect with that svabhâva is not in the con
ditions. Bhavaviveka likewise understands 3ab as s vat a utpadapratisedha (see Pan
deya 25).
20 Garfield (1995: 11 If) gives a superficially similar but far more sophisticated
reading. On his view, 3ab is meant to show that anything originating from conditions
must lack intrinsic nature. 3cd then seeks to refute origination from distinct things by
drawing the consequence from 3ab that nothing has intrinsic nature, hence that
nothing can be said to be truly distinct from other things. Things cannot be said to
originate from what is other because, lacking intrinsic natures and thus ultimate
reality, things cannot be ultimately distinct from other things. The chief problem I
see in this reading is that on it, Nâgârjuna still lacks an argument for the claim that
anything that is causally dependent must lack intrinsic nature, so that Hayes' charge
still stands.
21 Candraklrti actually gives two interpretations of 3cd. The first is the one about to
be discussed. The second employs something like Taber's principle of coexisting
counterparts: tat kim apeksam paratvam pratyayâdinâm, vidyamânayor eva hi mai
tropagràhakayoh parasparâpaksam paratvam, na caivam bljânkurayor yaugapadyam/
tasmàd avidyamâne svabhâve kâryanâm parabhâvah paratvam bljâdlnâm nâsàti
paravyapadesâbhâvâd eva na parata utpâda iti (de la Vallée Poussin 78, Pandeya 26).
'With reference to what is there that otherness of conditions etc. [when the effect has
not yet arisen]? Just as there is found mutual dependence of Maitra and his master.
Therefore intrinsic nature not being found, the extrinsic nature of the effects is not
the otherness of the seeds, etc., due to lack of designation of the other, origination is
not from what is distinct'.
22 This is how Candrakirti explains the introduction of the notion of kriyâ at MMK
1.4. He says that since the opponent is unable to explain the production of con
sciousness just on the basis of a distinct set of causes and conditions having existed in
the prior moment, they will claim these give rise to a consciousness-producing action
or causal force, which in turn produces the effect of consciousness, tad evam prat
yayebhya utpâdavâdini pratisiddhe, kriyâta utpâdavâdimanyate, nacaksütüpádayah
pratyayâh sâksâd vijnânam janayanti, vijnânajanikriyânispàdakatvàt tu pratyayâ
ucyanté / sa ca kriyâ vijnânam janayati / tasmât pratyayavatâ vijnànajanikriyâ
vijnânajanikâ na pratyayâh/yathâ parikriyâ odanasyeti (de la Vallée Poussin 79,
Pandeya 26).
23 The Nyâya version of asatkâryavâda involves synchronic causation. It is intro
duced to explain their claim that when a certain set of simple substances (e.g., the
atoms in a lump of clay) is arranged in a certain way, a new substance (e.g., a pot)
comes into existence. Naiyâyikas hold that the atoms and the pot are then related
through inherence (samavâya) as cause and effect. See Shaw (2002: 214f).
Abhidharmikas, as mereological reductionists, deny that the pot is ultimately real
and thus that the pot is the effect of the atoms. Their version of asatkâryavâda
involves diachronic causation; hence the appeal for them of the notion of a kriyâ or
causal force.
24 Garfield (1995: 113) also interprets the verse this way.
25 The verse itself is clearly open to this interpretation. In addition, this argument is
at least suggested by the Akutobhayâ comments on MMK 1.4:
That which is called by you an action pertaining to existents with the four conditions
is considered either to have conditions itself or not to have them. Whether it is

81.38.69.71 on Thu, 31 Mar 2022 09:46:50 UTC
existent or non-existent or both, since the conditions are in

does not exist. And the conditions are thought to have an a
found among them whether they are existent or non-existe
The positive example cited by the author in support of the

however, that of the burnt and the unburnt (kriyàvad va d
this in turn suggests the argument CandrakTrti mentions i
10.13:
dagdham na dahyate tavad adagdham naiva dahyate /

dagdhadagdha vinir muktam dahyamanam na dahyate //
This argument, CandrakTrti makes clear, follows the same reasoning pattern as does
the argument of MMK 2 concerning where motion takes place. And that argument
can be understood as employing the logic of the three times (see de la Vallée Poussin
21 If, Pandeya 199)) Thus the example of the burnt and unburnt employed in
Akutobhayâ could be taken as showing that the argument of MMK 1.4 is actually a
version of the same 'three times' argument deployed in MMK 1.5-7.
26 See MMK 4.6cd: na kâranasyâsadrsam karyam ity upapadyate, on which Cand
rakTrti comments laconically: bhinnalaksanatvàn nirvanavad evety abhiprayah (de la
Vallée Poussin 126, Pandeya 83). I take the point here to be just a refinement of a
Samkhyan argument against asatkàryavâda to the effect that if causation is dia
chronic, as with the lump of clay and the pot, then if the effect does not pre-exist in
its cause, why does an effect of just this sort (a pot) arise, and not any other (curds,
say, or a golden ring)? It is up to the asatkâryavâdin to provide some account of the
necessary connection we take there to be between what we consider cause and effect
such that upon the occurrence of sense-object contact, a sensory cognition will occur
and not nirvana. Perhaps resemblance between cause and effect might go some way
toward allaying this suspicion that on their account anything might follow anything
else. But where cause and effect are both distinct and do not resemble one another,
this suspicion remains in full force.
27 This is in effect how Naiyâyikas responds to the Advaitan infinite regress argu
ment against their formulation of asatkâryavâda. See Prasastapâda's discussion of
inherence in Padârthadharmasamgraha IX. 161.
28 If we follow Buddhapálita and CandrakTrti in their interpretations of verse 4,
there is no infinite regress argument in this verse. Instead, MMK 1.4—7 all present
variations on a single argument, that of the three times.
29 Strictly speaking it is MMK 1.7 that deals with the notion that the hetu is pro
ductive or originative in nature. Verses 5 and 6 simply assert that the alleged con
ditions cannot be designated as such either before or after the effect has arisen. But
CandrakTrti takes the argument of these verses to rely on the idea that in order for
some set of entities and events to count as hetu, they must collectively serve to bring
about or produce the effect. He comments on 6c: 'How would it be a condition of a
non-existent object, of something not found? If it were suggested that the designation
will come from the future existence [of the object], this is not so:
If designation is posited on the basis of future existence,

Not without power is there the coming to be of this'.
(de la Vallée Poussin 82, Pandeya 32)
And the difficulty with the notion of power (sakti) is precisely that it is no more than
a sort of black box that is posited in order to close the explanatory gap between cause
and effect. To attribute a power to the conditions is to say no more than that they are

81.38.69.71 on Thu, 31 Mar 2022 09:46:50 UTC
418 MARK SIDERITS
able to bring the effect into existence; it i

alone explain how the production can occ
of why these conditions should be followed
have the power to is to say no more than
30 Note that there are a host of sorites d
partite entity of a certain complexity, the
neither that the entity exists nor that it
that a real entity might neither exist no
entities must be conceptual construction
31 It might be thought that strict simultan
might precede the effect in origination pr
the two. But in that case it must be asked
the initial duration of its existence. If n
again faced with a mystery. But if some
necessary for maturation of the conditions
called the cause is not truly the cause at
this additional factor. And now we may
the effect. This set of conditions must occur
effect has come into existence, or else t
makes this point in commenting on MM
santah kim cid anyam pratyayam apeksya
ayuktam / y at tatpratyayântaram apr
pratyayatve satyasya pratyayo bhavati / t
Vallée Poussin 82, Pandeya 31).
32 The standard answer is that a possib
propositions. This raises the further quest
are said to be eternal, a Buddhist would n
'third-realm' entities (neither physical ent
orthodox Indian philosophers. For an
Samvâda: A Dialogue between Two Philoso
M.P. Rege, R.C. Dwivedi and Mukund La
33 David Armstrong is largely responsible
was also used by Nyàya. Fales (1990) provid
34 In Chapters 7 and 8 respectively of S
35 The key divergence comes in our respec
takes as an argument that commits the fa
an existent cannot derive its intrinsic n
dependent (where causal dependence is no
Nàgàrjuna as here trying to establish th
of intrinsic nature. I see the verse as instead
effect is not found among the condition
distinct entities with distinct natures, it r
arise from the cause. Thus Nàgàrjuna is h
the asatkâryavâdin of explaining why on t
can arise from anything. But this concer
stood as obtaining between ultimately re
question whether the effect can itself be
REFERENCES
Burton, D. (1999). Emptiness Appraised'. A Critical Study of Nagarj

Richmond: Curzon.

81.38.69.71 on Thu, 31 Mar 2022 09:46:50 UTC
Fales, E. (1990). Causation and Universals. London: Routled

Garfield, J.L. (1994). Dependent Arising and the Emptiness
Nâgàrjuna start with causation? Philosophy East and Wes
Garfield, J.L. (1995). The Fundamental Wisdom of the Mid
Hayes, R.P. (1994). Nàgàrjuna's Appeal, Journal of Indian
Shaw, J.L. (2002). Causality: Samkhya, Bauddha and Nyây
Philosophy 30, 213-270.
Siderits, M. (2003). Personal Identity and Buddhist Philosop
dershot, UK: Ashgate.
Taber, J. (1998). On Nàgàrjuna's So-called Fallacies: A Com
Indo-Iranian Journal 41, 213-244.
Tillemans, T.J.F. (2001). Trying to be Fair to Mâdhyamika
Yehan Lecture in Buddhism, Winter 2001, Calgary, Alberta:
ABBREVIATIONS
AKBh Abhidharmakosabhâsyam of Vasubandhu, ed. Prahlad Pradhan.

Patna: Jayaswal Research Institute, 1975.
MMK Mûlamadhyamakakârikâ, ed. Raghunath Pandeya as: The Mad
hyamakasâstram of Nâgârjuna, with the Commentaries Akutob
hayâ. by Nâgârjuna, Madhyamakavrtti by Buddhapálita, Prajñ
âpradïpavrtti by Bhâvaviveka, and Prasannapada by Candrakïrti.
Delhi: Motilal Banarsidass, 1988.
Mûlamadhyamakakârikâs (Mâdhyamikaswtras) de Nâgârjuna
avec la Prasannapdâ Commentaire de Candrakïrti, ed. Louis de la
Vallée Poussin. Osnabrück: Biblio Verlag, 1970.
VM Visuddhimagga of Buddhaghosâcariya, ed. Henry Clarke Warren,
rev. by Dharmananda Kosambi. Cambridge: Harvard University
Press, 1950.
Department of Philosophy
University of Liverpool
Liverpool L69 7WY
UK
E-mail: msideri@ilstu.edu

81.38.69.71 on Thu, 31 Mar 2022 09:46:50 UTC

Causation and Emptiness in Early Madhyamika

Uploaded by

Copyright:

Available Formats

Causation and Emptiness in Early Madhyamika

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Causation and Emptiness in Early Madhyamika

Uploaded by

Copyright:

Available Formats

CAUSATION AND EMPTINESS IN EARLY MADHYAMAKA

Author(s): MARK SIDERITS

Stable URL: https://www.jstor.org/stable/23496964

This content downloaded from

CAUSATION AND EMPTINESS IN EARLY MADHYAMAKA

It has long been recognized that there are important c

Journal of Indian Philosophy 32: 393-419, 2004.

This content downloaded from

A central Abhidharma project was to make philosophical sense of the

This content downloaded from

real. This led the Àbhidharmikas to search for the genuinely

This content downloaded from

meant anything like 'self-existence

This content downloaded from

agree with dualists and physicalists about is just whether thing

This content downloaded from

to think of it as just the occurrence of liquidity. Then wha

This content downloaded from

real - it will disappear upon complete analysis. To put the

This content downloaded from

ultimate entities grasped by the five external senses) is

This content downloaded from

object contact is of a completely different kind than the material

This content downloaded from

2. There are four conditions: the primary cause, the o

In verse 1 Nâgàrjuna lays out the overall strategy of the arg

This content downloaded from

is this not true, it is clearly absurd. Typically we want to k

This content downloaded from

The classical commentators suggest another way of con

This content downloaded from

make at this point is to invoke a causal relation to do the

This content downloaded from

some account of why we are justified in supposing that gi

This content downloaded from

This content downloaded from

produce a non-existent effect than a real person can be ador

This content downloaded from

This content downloaded from

involved here. It is clearly not logical necessity: Hum

Nàgàrjuna has, then, some interesting and powerful arguments

This content downloaded from

MMK 1 as a whole to be an argument for an affirmative answer to

P: If a relational tie is conceptually constructed, then any property

The argument would thus be that since the causal connection is

This content downloaded from

constructed relational tie 'being sacred to' should not

This content downloaded from

nature mythically (that is, the statement could be taken as

1 Taber (1998: 236f) is an exception. My thanks to an anonymous refer

This content downloaded from

5 Strictly speaking this is not quite right.

This content downloaded from

view that dharma and svabhâva are identical, in his extended

This content downloaded from

19 Candraklrti begins his comment on

This content downloaded from

existent or non-existent or both, since the conditions are in

The positive example cited by the author in support of the

dagdham na dahyate tavad adagdham naiva dahyate /