How do Mdhyamikas Think?

Notes on Jay Garfield,

Graham Priest and Paraconsistency

Several philosophers and Buddhist Studies specialists have taken up

the question of where or whether the philosophers of the Middle Way
school, i.e., Madhyamaka, use some form of deviant logic, or a logic
which would not recognize fundamental theorems such as the law of
double negation, the law of excluded middle and even the law of noncontradiction. Often this investigation is focused on the tetralemma
(catukoi), with very varied results. In recent writings problems about
excluded middle or the interpretation of negation in the tetralemma
have tended to take precedence over earlier interests about
Madhyamaka respect or non-respect for the law of non-contradiction.
Indeed it's probably fair to say that, in Buddhist Studies at least,
attributing contradictions to the Ngrjuna has increasingly fallen out
of vogue, such an attribution or tolerance often being considered, by
those of a philosophical bent, as tantamount to a disastrous
trivialization of Madhyamaka as exclusively mystical or even
irrational. Some would argue intuitively that contradictions are
rationally unthinkable. Others would invoke a more sophisticated
problem that anything and everything would follow from a
contradiction so that all reasoning would become indiscriminate
contradictions thus could supposedly never be tolerated by rational
individuals on pain of logical anarchy. In any case, the underlying
idea then is that if the Madhyamaka is not to be trivialized as irrational
and indeed I agree it should not be so trivialized , it would have to
rigorously respect the law of non-contradiction.
Recently the Australian logician, Graham Priest has teamed up
with the American philosopher, Jay Garfield, to significantly elaborate
upon certain ideas that they attribute to the second century C.E.
author, Ngrjuna. To do this they rely on paraconsistent logic, i.e.,
formalizations of logic that eliminate the spectre of anarchy by not
allowing everything to follow from a contradiction. Amongst the
several types of such paraconsistent logics developed in the current
technical literature, Priest and Garfield opt for a radical type and
maintain that there are some Ngrjunian arguments that can best be

interpreted as evidence of a logic that allows an acceptance of some

true contradictions. In other words, Ngrjuna was (at least implicitly
or in a reconstruction of his philosophy) an advocate of "dialetheism",
i.e., the position that some contradictions are truefor some statement
, & not- is true. Priest and Garfield's joint paper, "Ngrjuna and
the limits of thought," certainly does not endorse a laissez-faire
acquiescence in any and all contradictions; they do seek to argue that
some statements of contradictions in Madhyamaka can be best taken
as true, notably those along the lines of "the ultimate truth is that there
is no ultimate truth" or "all things have one nature, i.e., no nature".
These and other logically similar statements that are supposedly to be
found in Western thinkers like Kant, Wittgenstein and Hegel, have the
characteristic that Graham Priest has diagnosed as being at the "limits
of thought", in that they involve totality paradoxes. The dialetheism
comes in when we say that specific sorts of totalities, or "inclosures",
exist and that there are at least some things which both are and are not
in them.1
My own take: I can readily accept a limited type of paraconsistency
in the Prajpramit and Ngrjuna, but full-fledged dialetheism
seems to me unlikely
Graham Priest and Jay Garfield seem to have read Tillemans (1999)
as, like them, accepting Ngrjuna's "sincere endorsement of
contradictions". Well, no doubt, finding out what that work`s svamata
(rang lugs, own take) on paraconsistency might actually be is a
difficult task, especially as my views evolved considerably since an
earlier article. It's probably by now high time to set out as clearly as I
can what I do accept.
In my introduction to Scripture, Logic, Language I had said:
"I don't now know how to exclude that the
Prajpramitstras are most simply and naturally read as
having more or less the contradictions they appear to have.
Indeed, that [Edward] Conze-[Jacques] May scenario
fascinates me more and more."2

I was essentially imagining an attempt at a more or less literal and

unhedged interpretation of certain passages in the Prajpramitstras and in early Madhyamaka writers like Ngrjuna, an
interpretation which would be independent of and even opposed to
that of the later commentators. I found myself in a position where I
could no longer rule out an interpretation of this sort on formal
grounds or because all paraconsistency supposedly would be irrational
or lead to unlimited anarchical implication. Indeed the prospect of
trying to tread this paraconsistent path seemed to me worthwhile, even
heady, in that it seemed to be an attempt to take the provocative and
disturbing aspects of the Madhyamaka seriously, straight-no-chaser,
and not explain them way with sophisticated ad hoc solutions or even
massive additions to the texts designed to accomodate a type of
prescriptive "common sense" about what was needed so that an author
like Ngrjuna would supposedly be minimally rational.
In Scripture, Language and Logic, I had spoken about a natural
and simple literal reading of passages in the Prajpramitstras
that suggested acceptance of contradiction. I was thinking primarily of
what can be called the "signature formulae" of the Vajracchedikprajpramitstra.3 These are the oft-repeated statements
throughout the stra that say that X does not exist or is not the case
and that we therefore say that X does/is. E.g., the Buddha doesn't have
any distinctive marks and that is why one says he does.
Vajracchedik p. 24 (ed. M. Muller):
buddhadharm buddhadharm iti subhte 'buddhadharm caiva te
tathgatena bhit / tenocyate buddhadharm iti // "'The dharmas
special to a buddha, the dharmas special to a buddha,' these the
Tathgata has taught to not in fact be/have dharmas special to a
buddha. Thus they are said to be dharmas special to a buddha."
Leave aside the somewhat tricky question as to how we take the
Sanskrit compound, abuddhadharm ca (as a bahuvrhi or as a
tatpurua I tend to opt for the latter, as do most translations in
Buddhist canons). In any case the simple and natural reading of the
passage, deliberately neglecting commentaries, is to say that this
signature formula is denying something and then later affirming it.
We'll go into such statements, as well as the commentaries, in more
detail below.

As for Ngrjuna, I was only secondarily thinking of his use of

the tetralemma: what impressed me was a natural, more or less literal
interpretation of his system in the sixfold logical corpus (rigs tshog
drug). At some points (e.g., in his Ratnval) he endorses various
Buddhist doctrinal positions (e.g., karmic retribution etc.) and at other
points (i.e., in the Mlamadhyamakakriks) clearly denies that there
are any such things at all. The fabric of his system, again neglecting
commentaries, seems to suggest contradictoriness not unlike that to be
found in the signature formulae of the Vajracchedik: such and such
is said to be so in certain texts, chapters, etc. and elsewhere, or even in
the same paragraph, is said not to be so.
My own mentor, Jacques May, had already many years ago
applied some Hegelian notions to Ngrjuna and had seen Ngrjuna's
dialectic as a continuous and endless Aufhebung (sublation) of
provisional contradictory positions. Edward Conze had also held a
view on the Prajpramit as accepting contradictions. Although I
couldn't see myself furthering the exact same perspectives that they
were promoting, I did think that the debate about paraconsistent
approaches, a debate that had gone out of fashion in Buddhist Studies,
should be profitably revived. That has now happened with Graham
Priest and Jay Garfield and it is, I think, something of a liberating
experience to be able to talk seriously about Madhyamaka and
Prajpramit in these terms, even if, as we shall see, I don't share
their enthusiasm for dialetheism in Buddhism.
Priest and Garfield on Ngrjuna
Although a natural account of the signature forumulae of the
Vajracchedik and the fabric of Ngrjuna's six works seems to go
have him endorsing the truth of at some point and endorsing the
truth of not- at another, we never, to my knowledge, find a clear
textual endorsement of the truth of the conjunction, and not-. My
intuitive skepticism about dialetheism in Ngrjuna is thus obviously
going to clash with Graham Priest's and Jay Garfield's interpretation.
We thus need look in more detail at why Priest and Garfield thought
that there were indeed true contradictions (i.e. true conjunctions of
and not- ) in the Madhyamaka and what exactly would be
problematic in such a reading of Ngrjuna.

Let's be clear on a point of method: attributing such and such a

type of logic to an author like Ngrjuna is primarily a matter of
rational reconstruction. If we are going to cite texts, and I certainly
think we should, the onus has to be heavily on their interpretation,
context and rational reconstruction the mere words by themselves
prove virtually nothing. So far, I think, Priest, Garfield and I would
agree, and I would have no problems with their characterization that
what they are doing with Ngrjuna "is .. not textual history but
rational reconstruction."4 Indeed, if we look at the two candidates for
paradox that they give, these statements are certainly not quotations
from Ngrjuna. The first one, i.e., "The ultimate truth is that there is
no ultimate truth", is in effect arguably a consequence or paraphrase of
several passages in the Mlamadhyamakakriks and in that way can
perhaps be claimed to be the thrust of Ngrjuna's philosophy, if not
actually his words.5 The second, i.e., "all things have one nature, i.e.,
no nature", is in fact close to an historically attested interpretation of
some passages in Madhyamakakriks XV about the three
characteristics of any intrinsic nature: non-fabricated, independent of
other things (nirapeka paratra) and always fixed. It is especially
Candrakrti's interpretation of these passagesnotably his
Prasannapad on XV.2that brings in the idea that Ngrjuna does
not just refute intrinsic natures, but accepts that there is at least one
such non-fabricated, independent and unchanging fixed nature of
things, viz., their emptiness (nyat).6
I shouldn't go into very many technical details here about how
Priest and Garfield formally present the paradoxes that they see in
Ngrjuna; I'll confine the presentation of inclosure schemata to a long
footnote. 7 In any case, Priest and Garfield claim that the Ngrjunian
paradox is one like set theoretical paradoxes, where in the case of
some totality, or "limit of thought", defined in a certain way, there will
be objects that both are included within it and are outside it. They
argue that if we suppose with Ngrjuna that all things are empty, then
the totality of empty, i.e., natureless, things will itself have a nature
(i.e., being empty) and that this nature will be both in and not in the
totality of empty things. Formulated as an inclosure paradox it does
look potentially interestingly similar to other such paradoxes, like
those of Cantor and Russell.
The connection with other logical paradoxes would itself be
quite important as it would serve in part to answer a charge that these
Ngrjunian paradoxes are simply "rhetorical paradoxes" along the

lines of "The Golden Rule is that there is no Golden Rule", "I can
resist everything except temptation" and other such cute duplicitous
sayings that are no more than attention grabbers.8 Undeniably there is
a penchant for such enigmatic, provocative styles of expression in
Indian philosophy, so that it would be silly to say that every use of
words in an apparently contradictory fashion in a Sanskrit text is a
case of an author embracing dialetheism or saying something exotic of
logical interest.
The saving grace of Priest's and Garfield's
Ngrjunian paradoxes, if they are right, would be that they would
both be cases of a wider East-West phenomenon, i.e., inclosure
paradoxes, and would thus not be simply a matter of provocative style.
They would supposedly be logical paradoxes in the same way that
Russell's paradox and Cantor's are.
Although the Ngrjunian paradoxes would be interesting for
comparative philosophers in that they would be East-West discoveries
of consistency problems with totalities, I see two major problems in
seeing totality paradoxes as playing a significant role in Ngrjuna's
thought, i.e., philosophical problems and historico-textual problems.
Let's take philosophical considerations first. One problem is
that these paradoxes would only happen in very specific areas: the
biggest sets, the widest applicable properties (like emptiness), the
limits of conceivability and expressibility, the limits of thought.
Given this exclusively "big picture" in which Ngrjuna would figure
with his fellow explorers of the transconsistent, it is hard to see that
his paradox and the supposedly resultant dialetheism is going to have
much bearing on the basic fabric of his Madhyamaka philosophical
system, such as his idea of the two truths, their inseparable
connection or the identity of sasra and nirva. We are left a bit in
the lurch as to how to interpret his more basic and pervasive doctrines
other than the rather isolated totality paradoxes.
In fact, it looks to me that there may be something of a nonsequitur in what Priest and Garfield claim the Ngrjunian inclosure
paradoxes show. They claim that the second paradox about emptiness
shows the following:
"All phenomena, Ngrjuna argues, are empty, and so
ultimately have no nature. But emptiness is, therefore, the
ultimate nature of things. So they both have and lack an
ultimate nature."9

This conclusion seems to involve a certain leap, in that Priest and

Garfield are using the big picture paradoxes to come up with a view
on each and every individual thing. I'm not at all sure that this works.
After all, the point about the emptiness paradox is that one
very important universal property, emptiness, both is and is not empty,
or in other words, lacks an intrinsic nature and does not lack one. This
paradox concerns the nature of the totality of empty things and it is
that nature that is shown to be both empty and not empty. Indeed, that
is the way the emptiness paradox is formulated by Priest and Garfield:
"The limit-contradiction is that the nature of all things ()
viz., emptinessboth is and is not empty.
But immediately after this, they then say:
Or to repeat Ngrjuna quoting the Prajpramitstra, 'all
things have one nature, that is no nature.' "10
Now, whether individual things, like tables and chairs, have
the contradictory properties of being both empty and not empty looks
like a separate matter from whether emptiness itself has those
contradictory properties. Indeed all things, like tables, chairs, etc.,
having no nature and yet also having as their nature emptiness would
yield a paradox of them having and not having a nature, i.e., being
empty and not empty. That may perhaps be where Ngrjuna and the
author(s) of the Prajpramitstra want to go in saying that all
things have one nature, i.e., no nature. But it's not obvious that they
get there from the Ngrjunian paradox that Priest and Garfield have
formalized; that remains, after all, a "limit-paradox" about emptiness
itself and does not clearly show that each table and chair have the
same contradictory properties that emptiness itself supposedly does. It
doesn't seem to follow that if there was a inclusion paradox about a
universal set , or an intrinsic nature of , the things in would
themselves exhibit the same paradoxicalness.11
Secondly, historical and textual considerations. It seems to me
difficult to account for the spirit and letter of Indian texts and also say
that Ngrjuna would advocate true contradictions that he would see
as stemming from totality paradoxes. Consider what we do know
about the "spirit" of Indian discussions of totalities. Indeed,
arguments about inconsistency in the notion of a totality do explicitly

and repeatedly figure in Indian philosophy's arguments about the

coherence of the notion of sarva (all, the totality, the universe), but
they are certainly not accepted as cases of dialetheism or genuine true
contradictions. They are instead used by Naiyyikas, like
Uddyotakara in his Nyyavrttika to Nyyastra 2.2.6612, to refute
Buddhist doctrines, like the semantic theory of apoha, by reductio ad
absurdum. The non-Buddhist Naiyyika argues, for example, that the
Buddhist theory of apoha, where the term X signifies non-non-X, is
impossible in the case of word like "the totality" "all" (sarva). The
reason is that the Buddhist would (absurdly) have to admit that there
is something, be it a set or a property or an individual, that is outside
the totality of things, because given the principal that any word X
signifies non-non-X, "the totality" would express the negation of nontotality (asarvanivtti) the hidden premise is that there always must
be something to negate, a real, existing negandum. Of course the
problem then is that this something outside the totality would have to
be both outside and not outside the totality of things.
Whatever the value of Uddyotakara's attack, the Buddhists,
including Madhyamakas like Kamalala in his Madhyamakloka and
ntarakita in the apoha chapter of his Tattvasamgraha, repeatedly
take great pains to show that these totality paradoxes are not
paradoxes at all, and that there is a way to preserve consistency by
saying that "something outside the totality" (asarva) is just a
conceptual invention: it's not necessary that it actually be real to be
the negandum in non-non-totality. In short, the most explicit and
frequent discussions about contradictions stemming from totalities are
those between non-Buddhists and Buddhists; the Buddhists defend
themselves by arguing that acceptance of totalities does not lead to
any inconsistencies at all. The Buddhists never, as far as I know,
accept any explicitly formulated argument, non-Buddhist or
otherwise, to the effect that sarva, the universe, the totality, would be
contradictory. It would thus be odd that they would somehow
implicitly promote true contradictions involving totalities elsewhere
in their philosophy.
Turning to the letter of the texts, I would argue that not just are
totality arguments used differently in Indian philosophy from the way
Priest and Garfield would have it, but that it seems rather implausible
to say that Ngrjuna himself accepted any true contradictions at all:
indeed he seems to give pretty good textual evidence that he does not.

For example, in Madhyamakakrik XXV.14 he gives what looks

like a clear prohibition against contradiction:
bhaved abhvo bhva ca nirva ubhaya katham /
na tayor ekatrstitvam lokatamasor yath //
"How could both non-being and being pertain to nirvna? Both are
not present in one place, just as light and darkness [are not present in
one place]"
Even more explicit in banning true contradiction is Candrakrti's
comment:
bhvbhvayor api parasparaviruddhayor ekatra nirve nsti
sabhava iti // bhaved abhvo bhva ca nirva ubhaya katham /
naiva bhaved ity abhiprya / "For being and non-being too, there is
no possibility for the two mutually contradictory things
(parasparaviruddha) to be present in one place, i.e., nirva. Thus
"how could both non-being and being pertain to nirva?" The point
is they could not at all.
The argument is situated in the context of the fourfold negation of the
tetralemma (catukoi), where an opponent suggests that nirva both
is and is not in short the opponent is advocating that a true
contradiction would apply. Ngrjuna and Candrakrti reply that such
a true contradiction is not possible. Now, there is no indication
whatsoever that their reasoning is restricted to some isolated specific
case, i.e., nirva. It looks pretty clearly generalizable: no true
contradiction in this case, because no true contradictions at all. To
suggest otherwise and say that there are some true contradictions in
certain specific cases, makes for a circumscribed rejection of the third
lemma of the tetralemma that is hard to reconcile with fundamental
Madhyamaka texts. Indeed, there is solid evidence (e.g., in the work
of Ngrjuna's disciple, ryadeva, in Candrakrti and in others) that
the essence of the Madhyamaka method is that the rejection of the
four lemmas in the tetralemma is and must be generalizable and is the
way a Mdhyamika should always proceed in criticizing philosophical
positions. This is the point of the famous verse in ryadeva's
Catuataka XIV.22 which advocates negating all four positions:

sad asat sadasac ceti sadasan neti ca krama / ea prayojyo

vidvadbhir ekatvdiu nityaa // "Being, non-being, [both] being and
non-being, neither being nor non-being: such is the method that the
wise should always use with regard to identity and all other [theses]."
Again, it would be quite odd if a Buddhist were to allow that, inspite
of verses like this, Ngrjunian totality paradoxes nonetheless yielded
true contradictions that were "exceptions" to the rejection of the third
lemma.

How to read Ngrjuna and the Prajpramit naturally and yet

not embrace dialetheism
Let me try to make the best case I can for a paraconsistent
interpretation of Ngrjuna and the Prajpramitstras. Instead of
"totality arguments" suggesting some form of paraconsistency or even
dialetheism, as Priest and Garfield have it, I think that it is the whole
system that suggests a type of paraconsistency on a natural reading,
and in a particular it is the use of the two truths, customary
(savtisatya) and ultimate (paramrthasatya). I'll take up the matter
of which form of paraconsistency below, but suffice it to stress again
that it is not full-fledged dialetheism.
Here's a key Indo-Tibetan problem of interpretation when
dealing with the two truths in early Madhyamaka and Prajpramit
texts. Suppose the Buddhist author(s) says or somehow endorses that
is true and also say that not- is also true, as the Mdhyamikas and
the author(s) of the Prajpramitstras are wont to do when they
say that dharmas, aggregates, Buddha-marks, karma, suffering, etc.
exist and also say that they do not exist, i.e., are empty. Are
parameters, like the qualifiers "customarily" and "ultimately" implicit,
somehow built-in to and not- respectively so that there is only a
pseudo-appearance of contradiction?13 Or is it the same statement,
without any implicit parameters, whose truth is being endorsed (for
one set of reasons) at some points in the text and rejected (for another
set of reasons) at other points in the text?
By way of illustration of the two approaches, let us go back to
the signature formulae of the Vajracchedikprajpramitstra: X
does not exist/is not the case, and thus we say that X does exist/is the
case. E.g., the Buddha doesn't have any distinctive marks and that is
why one says he does. Now these types of statements can be and have
been approached in both above-mentioned ways. We could do what
Kamalala did in his Vajracchedikk, and which was a kind of
common later Buddhist interpretative strategem, namely, clearly
differentiate the perspectives involved: the buddhadharmas, etc. are
not buddhadharmas,etc., looked at ultimately, and are buddhadharmas,
and so forth looked at from the point of view of customary truth.14 We

could either add explicit qualifiers right into the wording of the
respective affirmative and negative statements (as did for example the
Tibetan writer Tsong kha pa (1357-1419)) or we could leave the
actual wording in the stra unchanged but say that qualifications of
perspective had to understood, as did the 8th century Indian
Mdhyamika, Kamalala. In any case, for our purposes, the result is
more or less the same: appearance of contradiction vanishes. There
would be nothing more logically provocative here in endorsing both
statements than there would be in endorsing the statements "It is ten
o'clock" and "It is not ten o'oclock", when we also know that the first
statement concerns Eastern Standard Time and the second concerns
Pacific Standard Time.
We could also adopt an approach that leaves the provocation of
the signature formulae intact, i.e., say that the same completely
unparameterized statement is being affirmed and negated. In short, no
explicit parameters nor timezone-like switches of perspective along
the lines of Tsong kha pa or Kamalala, but only radically different
kinds of supportive reasoning as to why the one statement is true and
why its denial is true. Thus, the stra author(s) would have good
reasons to say that dharmas or Buddha marks do exist (e.g., to account
for truths that must be accepted in the world, or at least amongst
worldling Buddhists) and other good reasons to say they don't (e.g., to
give an account of their ultimate status, emptiness). Indeed this move
might well bring out just how close and inseparable customary and
ultimate truths are for early Mahyna Buddhist authors: as
Ngrjuna had himself oft repeated, the customary, i.e., sasra, is
nothing but (eva) the ultimate, i.e., nirva, and vice versa: put in our
terms, the two truths are so close that the very same unparamaterized
statements about dharmas, aggregates, suffering, etc., are both
asserted and denied.
There is, I think, a good reason to prefer the second style of
interpretation and be suspicious about the imposition of hedges and
parameters and other attemps at non-literal nuancing of early
Madhyamaka. Simply put: If we can read Ngrjuna and the
Prajpramit without qualifiers and pretty much literally, let's do it:
Ngrjuna is interesting and intelligent as is and charity will not
require anything more.
A not infrequent, but significantly less persuasive, argument is
that a nuanced and finessed approach is downright wrong, or even a
travesty to Madhyamaka, because it would bring in philosophical

theses by the back door and thus fatally weaken the Madhyamaka
project.15 Indeed there is a traditional Sa skya pa interpretation of
Madhyamaka that goes a considerable distance in fleshing out this
argument: I'm thinking of the lTa ba'i shan 'byed of the fifteenth
century Tibetan writer, Go rams pa bSod nams Seng ge (1429-1489),
probably the most explicit traditional source I know of for a
potentially coherent, unparamaterized interpretation of Ngrjuna and
the Prajpramit. Go rams pa's target was of course Tsong kha pa,
who advocated adding qualifiers like "ultimately" (don dam par), or
"truly" (bden par), to all the negative statements in the Madhyamaka:
things are not ultimately/truly produced, are not ultimately/truly
existent, and so on and so forth for all properties a person might wish
to attribute.
Go rams pa's main point, in his refutation of Tsong kha pa, was
that if a Mdhyamika commentator adds that kind of ultimate
parameter and thus gives a non-literal interpretation of the Ngrjuna's
negative statements, he has in effect denatured the whole Ngrjunian
dialectic to the degree that it will no longer be able to accomplish its
(religious) purpose of quietening philosophical speculation and
attachment and irenic quietism, or complete "freedom from
proliferations" (spros bral, niprapaca), is for Go rams pa the main
point of the negative dialectic of the Madhyamaka. So his alternative
is thus to take literally the idea of yod min med min "no existence, no
non-existence" and not add any qualifiers like "ultimately"/"truly", the
danger being that by negating "ultimately " instead of itself for
any statement, the Mdhyamika thinker will arrive at smugness about
being free of positions, but will in fact remain as attached to the truth
of as any other realist philosopher. Qualifiers and hedges, so the
argument can be paraphrased, make everything a little too neat and
refute straw men.
Let's suppose we adopted an unqualified reading either because
simplicity and naturalness are (all other things being equal) better
than artifice, or perhaps because of Go rams pa-style quasi-religious
arguments. Where does this unhedged interpretation of Ngrjuna and
the Prajpramit take us in terms of logic? I would say the
following: it leads to a type of paraconsistent logic according to which
Ngrjuna will in certain discussions admit that is true (for worldly,
doctrinal or even Abhidharmic reasons) and in other contexts that
is true (for reasons involving emptiness of intrinsic nature); however,
Ngrjuna will recognize no good reasons at all to ever admit the truth

of the conjunction, &. There will be no such reasons, because

Ngrjuna, as I had argued earlier, is deeply respectful of the third
negation in the tetralemma, a negation which he generalizes to apply
to every statement. In short, we might end up a type of
paraconsistency, but not dialetheism if the Madhyamaka is not to run
afoul of its own prohibitions. For Ngrjuna, there is no , such that
& is true.
Now endorsing and endorsing , but refusing the move to
&, does seem to involve a type of paraconsistency, indeed a
recognizable one, arguably significantly similar to what Nicholas
Rescher and Robert Brandom developed in their joint book, the Logic
of Inconsistency. 16 A number of years ago, in a note liminaire to a
Felicitation volume for Jacques May who was indeed an
uncomprising advocate of a Madhyamaka with no qualifiers or hedges
whatsoever I had mentioned that Rescher and Brandom's (weak)
inconsistency might allow us to rationally reconstruct aspects of a
May-style Madhyamaka. In fact, I later discovered that the approach
was not unique to Rescher and Brandom: it was, as Koji Tanaka
pointed out in his taxonomy of contemporary theories of
paraconsistency, initially developed by the Polish logician, Jakowski,
and certain other writers, including some of my Canadian compatriots.
Tanaka classified these theories as non-adjunctive approaches to
paraconsistency, i.e., ones that prohibit the move from individual
premises, , , to their adjunction & 17 In other words, it enables
one to affirm that at least in some cases is true and to affirm that
is true a weak inconsistency without, however, ever admitting the
truth of the statement, & this latter statement would be a strong
contradiction that could not be accepted as true in the RescherBrandom system if dastardly consequences like explosion were to be
avoided.
Finally, following this scenario the logic of Ngrjuna would
have a quite remarkable split-level construction. Non-adjunctive
paraconsistency would apply to discussions of emptiness, like in the
tetralemma where Buddhists endorse , , but do not endorse the
adjunction &. It would also, as we have arguing, apply in a
general manner to the so-called six logical works (rigs tshogs drug) of
Ngrjuna, which sometimes treat of things worldly and then deny
those same things elsewhere in discussions of nyavda.
Paraconsistency would, not however, be supposed in more ordinary
discussions of customary matters, like the usual and banal reasonings

about fires on smoky hills, sound being impermanent, fire being hot
and other such non-ultimate considerations. Nor would it be
applicable in the numerous reductio ad absurdum (prasaga)
arguments where Ngrjuna seeks to show inconsistencies in his
opponents' positions. These reductio usually proceed by deriving at
some point and at anotheradjunction of the two gives the needed
contradiction. In sum, we probably could say that so long as
Ngrjuna discusses only such customary things as dharmas,
aggegrates and so forth, or when he demolishes others' philosophical
views, his logic does stay quite classical. The non-adjunctive
paraconsistency would come in in two places (a) when Ngrjuna
discusses the ultimate status of things in tetralemma-like terms. (b)
When we seek to characterize Ngrjuna's own stance by putting
together the different unqualified statements in different chapters of
his works: the trick would be to keep all those statements unhedged,
but to avoid strong contradiction by blocking their adjunction. I think
that dialetheism, however, would probably not apply at all.
Final caveats and conclusions
I think this is about as far as I can go in making a case for
paraconsistency in early Madhyamaka and the Prajpramit. In any
case, later Madhyamaka in India or Tibet, or in other words
Ngrjuna's philosophy as viewed by commentators from about the
sixth century C.E. on, is another story and is much more inclined to
paramaterization. It is also much more conservative about consistency.
There are more explicit prohibitions against virodha/viruddha
(contradiction), a fact which makes it more difficult to read the later
Mdhyamika scholastics as tolerating or advocating any
paraconsistency, be it weak or strong contradictions. Significant too is
that the later Mdhyamika is, with one or two exceptions, under the
spell of Dignga and Dharmakrti's logic so that there is an attempt to
harmonize Madhyamaka with a logic for realists. I'll give a few
examples in an appendix to illustrate how this is done with
parameterization, diambiguation and other relatively predictable
moves but suffice it to stress for the moment that true (strong or
weak) contradictions are anathema for the logicians of the DigngaDharmakrti school and there is every reason to think that they are too

for later Mdhyamikas, like Kamalala, ntarakita and many

others, who see themselves as under the same constraints as their
logician coreligionnists. There was a significant change in orientation
between Ngrjuna and later commentators (especially those of the
majority Svtantrika persuasion), and that change is largely due to the
overwhelming influence of the Dignga-Dharmakrti school on Indian
Buddhist thought.18
Would this evolution mean that later Indian or Tibetan
Madhyamaka is inauthentic or without philosophical value as a
development of Madhyamaka thought? Of course not, unless
authenticity demanded no evolution, a kind of pure doctrinal
deepfreeze. I think that we can and should defend Tsong kha pa and
others from Go rams pa's critique of grotesquely denaturing
Madhyamaka thought and yet we can also grant that Go rams pa was
quite right at least in saying that what Tsong kha pa did in massively
adding qualifiers to early Indian texts and in adopting Dharmakrti's
logic did not accord well with the logical structure of early
Madhyamaka thought.
We can advance the following historical claim: not only did
the philosophical debates and doctrines evolve over time with
Buddhist scholastic thinkers, but very possibly the logic evolved away
from its rather complex split-level architecture to one of increasing
homogeneity and classicalness. Logical simplification also happened
elsewhere in Indian thought. A number of us have looked at, for
example, the logicians' problems in explaining a theory of valid
reasons: with time the formal aspects became ever simpler, even
though the philosophical analyses often became increasingly subtle.
Something interestingly similar may well have happened with the
Madhyamaka.

Appendix:
Textual sources illustrating later Indo-Tibetan
approaches to inconsistencies.
I would argue that the typical later Madhyamaka approaches to
(seeming) inconsistencies (strong or weak) are quite recognizable ones
that are common to East and West: add parameters, diagnose
equivocations and ambiguities and in so doing the "inconsistency" is
defused.
An example. There is a provocative canonical statement in the
rya Dharmasagti that has led to considerable exegetical
manoeuvers on the part of later Mdhyamikas:
"Not seeing any dharmas, this is genuine seeing."19
The statement sometimes has an even starker formulation, as we find
it Atia's Satyadvayvatra verse 7:
"Non-seeing is seeing."20
As history goes on we find commentators making increasingly
explicit and precise attempts to satisfactorily respond to seeming
inconsistencies in such canonical statements. Thus the eighth century
Indian commentator, Kamalala, in the Madhyamakloka, gives a
rather involved dogmatic explanation to show that the non-seeing of
any intrinsic natures is not just a mere absence of seeinga so-called
"non-implicative negation" (prasajyapratiedha)but implies a type
of seeing, i.e., seeing the emptiness or ultimate truth about things.21 In
short, non-seeing in this context implies seeing something else. The
term adaranam in the stra quotation, especially the negative particle
a, is thus disambiguated to mean non-seeing understood in a special
way, much in the same way as the word "not" in a statement like "my
coat is not blue" can be interpreted so that it is implied that my coat
has some other color. In Buddhist logical jargon, the word adaranam
is said to express here an implicative negation (paryudsa) in that it
also implies a positive fact, i.e., seeing. In this commentator's view,
then, the problem turns on there being two types of negation; if a
seeming paradox arises it is because of a systematic ambiguity in the
negative particle in Sanskrit.

Not surprisingly perhaps, for Tsong kha pa the opposition

between "seeing" and "non-seeing" is easily defused by pointing out a
type of parameterisation. We can reconstruct his argument as follows.
The property, seeing, may appear to be a monadic property (i.e, such
and such a person sees), but actually in the stra context it is dyadic
(i.e., such and such a person sees such and such an object). The
opposition is defused when we can show that non-seeing pertains to
one thing and seeing to another.
"As for the meaning of 'not seeing is the supreme seeing,' we
don't accept that seeing nothing at all is seeing, but instead, as
we had explained earlier, we establish that not seeing
conceptual proliferations (spros pa= prapaca) is the seeing of
what lacks conceptual proliferation. Thus we do not apply
seeing and not seeing to the same basis."22
In no way do we see and not see one and the same thing. Examples of
other sorts of disambiguation of seemingly paradoxical canonical
statements can be provided, but this will have to suffice.

For the technical details on what an inclosure is, see fn. 7. In G. Priest and J.
Garfield (2002).
2

See T. Tillemans (1999), p. 12.

The phrase "signature formula" I owe to Paul Harrison in his introduction to his
forthcoming new translation of the Vajracchedik.
4

Priest (2002), p. 251.

It's worth mentioning, however, that later Tibetan commentators would hedge this
statement to make a distinction between ultimate truth and what is ultimately
established (don dam par grub pa) so that we end up with the tamer principle that
the ultimate truth is the ultimate truth, but, like any other dharma, it is not ultimately
established.
6

See Prasannapad (ed. L. de la Valle Poussin), 264.12-265.2: atha keya

dharm dharmat? dharm svabhva. ko 'yam svabhva? prakti. k
ceya prakti? yeya nyat. keya nyat? naisvbhvyam. kim ida
naisvabhvyam? tathat. keya tathat? tathbhvo 'vikritva sadiva sthpit.

sarvadnutpda eva hy agnydn paranirapekatvd aktrimatvt svabhva ity

ucyate. Note that in this vein there is also an important passage from the
Aashasrikprajpramitstra quoted in the autocommentary to Ngrjuna's
Vigrahavyvartan. See Priest (2002), p. 266. On the senses of svabhva, see also
de Jong (1972).
7

Here briefly is how it goes. The key formal notion that they introduce is that of an
"inclosure ": a totality set is an inclosure if (1) its members have a property and
have a certain property ; and (2) there is a "diagonalising" function that assigns to
each subset x of whose members have a new object that is not in the subset x
but is still in . Applying that function to itself we get an object that both is
and is not a member of . Symbolically, here are Priest's conditions for an
inclosure:
={y:(y)} exists, and ()
For all x such that x is a subset of and (x),
(x) is not a member of x
(x) is a member of
See Priest (2002), p.134 for the inclosure schema. In the Ngrjunian inclosure
paradox, is the set of all empty things, i.e., things without intrinsic nature; these
things have the property , i.e., having a common intrinsic nature. The diagonalising
function assigns to the subset x the nature of the things in x. Since (x) is a
member of , it does not have an intrinsic nature. x however has the property and
thus consists of things which do have a common intrinsic nature. Therefore (x) is
not a member of x. Applying to itself we have the result that () is a member
of and () is not a member of . The nature that assigns to is emptiness.
Thus emptiness is empty of intrinsic nature and is not empty of intrinsic nature.
8

On the distinction between logical and rhetorical paradoxes, see p. 4 in N. Rescher

(2001). See Tillemans (1999), p. 195-196 on some examples of the stylistic
tendency in Indian philosophy to use seemingly paradoxical modes of expression.
Often, it is quite clear (e.g. by looking at auto-commentaries) that these are only
rhetorical paradoxes.
9

Priest (2002), p. 268. The first paradox, conveyed by the statement "the ultimate
truth is that there is no ultimate truth" is spelled out as follows (ibid p. 269): "the
claim that there are no ultimate truths, both is and is not an ultimate truth" Here too
the move to each and every individual thing both having and not having some
property would not be obvious.
10

11

Priest (2002), p. 268.

Perhaps the underlying idea in Priest and Garfield is that emptiness is the nature of
things and is both empty and not empty, and thus all the things, whose nature it is,
should be empty and not empty too. But this seems to involve some non-obvious

steps. Interestingly enough, Priest seems himself aware that he has not proved (and
indeed he does not want to prove) that tables, chairs, stars and so forth are
contradictory:
"I claim that reality is, in a certain sense, contradictory. I do not, of course
mean that the objects that constitute reality, like chairs and stars, are
contradictory. That would simply be a category mistake. What I mean is
that there are certain contradictory statements (propositions, sentences
take your pick) about limits, that are true." (Priest 2002, p. 295)
Subsequently, he grants that (under the inspiration of Ngrjuna) he is also taking an
ontological turn, and that he is concerned not just with statements, but with the
nature of reality, with what is: that nature is what he takes to be contradictory.
However, formulating the problem in this way, the move from the contradictoriness
of that nature to the contradictoriness of each individual thing will still be
problematic.
12

See Nyyadaranam (ed. Taranatha Nyayatarkatirtha, Amarendramohan

Tarkatirtha, Calcutta 1936), p. 687. Transl. on p. 1056-7 in Ganganatha Jha, The
Nyya-stras of Gautama with the Bhya of Vtsyyana and the Vrtika of
Uddyotakara, Delhi: Motilal Banarsidass, 19??.
13

Here is how Priest describes parameterisation: "The stratagem is to the effect that
when one meets an (at least prima facie) contradiction of the form P(a)!, one tries to
find some ambiguity in P, or some different respects, r1 and r2, in which something
may be P, and then to argue that a is P in one respect, P(r1,a), but not in the other, P(r2,a). For example, when faced with the apparent contradiction that it is both 2
PM and 10 PM, I disambiguate with respect to place, and resolve the contradiction
by noting that it is 2 PM in Cambridge and 10 PM in Brisbane." (Priest 2002, p. 151)
14

See for example Kamalala's discussion on the stra's formula concerning "heaps
of merit", i.e., bsod nams kyi phung po, in ryaprajpramitvajracchedikk
(ed. Central Institute of Higher Tibetan Studies, Sarnath, 1994) p. 296-297.
15

Cf. Tillemans (1992), p. 11 on Jacques May's Madhyamaka philosophy: "Autre

lment clef: le bouddhiste Madhyamaka n'a pas, lui-mme, de position
philosophique. May est formel sur ce point et rejette les nuances ou les
affaiblissements de ce principe dans la littrature scolastique ultrieure. ... Le
Madhyamaka ne peut pas avoir de position philosophique, car toute position
s'annule."
16

I.e., Rescher and Brandom (1980). See Tillemans (1992).

17

See p. 29-30 in Tanaka (2003)

18

I have taken up the subject of the Svtantrika-Madhyamaka debt to the logicians

in "Metaphyics for Mdhyamikas." See Tillemans (2003).
19

P. 74b: bcom ldan 'das chos thams cad mi mthong ba ni yang dag par mthong
ba'o. Cf. the Sanskrit cited in ntideva's iksamuccaya p. 264, 1-2: adaranam
bhagavan sarvadharmn darana samyagdaranam iti. The passage has been
studied in detail in the third chapter of Keira (2004).
20

ma mthong ba nyid de mthong bar // shin tu zab pa'i mdo las gsungs. "The
extremely profound stra says that it is precisely non-seeing that is seeing." Text on
p. 194 of Lindtner (1981).
21

Here's a relevant passage from the Madhyamakloka (the translation is that of

Keira 2004, p. ???): "However, great yogins who correctly meditate on all dharmas
as being like mirages and echoes obtain the concentration (samdhi) which arises
from the ultimate limit of meditation upon the true state of affairs (bhtrtha), and
therefore they have inconceivable supreme powers.Thus this type of wisdom (ye
shes de lta bu), which clearly realizes (mngon sum du byed pa) the thusness of all
dharmas being selfless, arises in these yogins, so that (yena) the yogins understand
completely directly (mngon sum kho nar) that all dharmas are selfless. It is precisely
this that is termed nonseeing (adarana) [in the stra], but [the negation in
nonseeing] does not have the form of a non-implicative negation
(prasajyapratiedha). Madhyamakloka D. 168b-169a: on kyang rnal byor pa
chen po chos thams cad smig rgyu dang sgra brnyan lta bur ji lta ba bzhin du sgom
par byed pa rnams kyis / yang dag pai don bsgoms pai rab kyi mtha las byung
bai ting nge dzin thob pai phyir mthu phul du byung ba bsam gyis mi khyab pa
dang ldan pas rnal byor pa de dag la gang gis na rnal byor pa rnams kyis mngon
sum kho nar chos ma lus pa bdag med pa rtogs par gyur pa chos ma lus pa bdag
med pai de kho na nyid gsal rab tu mngon sum du byed pai ye shes de lta bu nye
bar skye ste / de nyid mthong ba med ces byai / med par dgag pai ngo bo ni ma yin
no //.
22

Tsong kha pa, dBu ma dgongs pa rab gsal (Sarnath edition) p. 202: mthong ba
med pa ni mthong ba dam pa'o zhes gsungs pa'i don yang / ci yang mi mthong ba
mthong bar mi bzhed kyi / sngar bshad pa ltar spros pa ma mthong ba ni spros bral
mthong bar 'jog pas / mthong ma mthong gzhi gcig la byed pa min no /.