and early sixth centuries. He is considered to have been a disciple of Vasubandhu, the author of
the Abhidharmakōsa.(1) At that time there was a great flourishing of philosophical thought both
within Buddhistm and among various schools of Hinduism, and real debate was engaged in by
contending thinkers. Dignāga was important in that he further developed a Buddhist approach to
epistemology and logic by which he could challenge the assertions of non-Buddhist
philosophers. His work was expanded on in the seventh century by Dharmakīrti, and together
their texts are the basis of all Buddhist logic practiced and studied today in the Tibetan scholastic
Though debate has, from the start, been the main way logic has been practiced, its true
purpose, rather than simply besting an opponent, is to learn to examine one’s own beliefs—to
subject one’s own presuppositions to rigorous analysis. Although Dignāga’s theories have been
criticized, both in his time and by modern scholars, his logic continues to be studied and valued
because it is such an effective method for working with conceptual mind, which is both the root
of all suffering, and the means to reach the understanding that dissolves suffering. In other
words, in Buddhist terms, it is a path. In this paper I will attempt to describe the main features of
Dignāga’s thinking and argue for its utility as a practice with a soteriological goal.
Dignāga’s Writings and Theories
Over 20 works attributed to Dignāga survive, mostly in Tibetan and Chinese translation.(2)
The most important of these are the Hetucakraḍamaru, in which he introduced a chart showing
all the possible combinations of the three modes of reasoning; and the Pramāṇasamuccaya,
which is considered to be his last and most important work.(3) The milieu in which Dignāga
operated was one of promiscuous dialogue and debate between and among Hindus and
Buddhists. One of the main areas of dispute among these various groups was the question of
what constituted pramāṇa, variously rendered as a “reliable warrant,”(4)“accredited source of
knowledge,”(5) or “valid cognition.”(6) Without some level of agreement on basic definitions, and
on what can be included as valid provenance for knowledge claims, it is impossible for debate to
be fruitful. Many of the Hindu schools that Dignāga criticized allowed for such things as
hearsay or scripture to be included as valid cognition. Dignāga allowed for only two: direct valid
cognition, or perception, and inferential valid cognition, or inference. He further said that these
two valid cognitions each have their own exclusive kinds of objects. For direct valid cognition,
the objects are particulars, or “specifically characterized phenomena,” and for inferential valid
cognition, its objects are generalities, or “generally characterized phenomena.” Another way to
say this is that for direct perception, i.e., the perception of a sense consciousness, its objects can
only be specific things, divorced from any generality imputed on them by conceptual mind
regarding its class, type, or quality, while the object of inference can only be mental constructs or
concepts with no actual reality. For Dignāga, the problem arises when conceptual mind mixes
the actual phenomenon with its name, conceptual class or category, and takes that mixture to be
the actual thing itself.
In addition to limiting the sources of valid cognition, Dignāga also insisted on the
presence of the three modes of reasoning in any syllogism or knowledge statement. This method
of analyzing arguments was already widely used before Dignāga, but his systematic presentation
of all the possible permutations of the three modes in his Hetucakraḍamaru was highly
influential. Any syllogism necessarily has three parts, the subject, predicate, and reason. For example, in the statement “sound is impermanent because it is produced,” “sound” is the
subject, “is impermanent” is the predicate, and “because it is produced” is the reason. In order to
test this statement for logical soundness, it has to be examined for the presence of three modes:
the subject quality, forward pervasion, and reverse pervasion.
The definition of the subject quality is “a reason that valid cognition has determined to be
present in all instances of the flawless subject in question, in a corresponding formulation.”(7) This
means that the reason, in this case being “produced,” has to be the present in all instances of the
subject, “sound.” In other words, its essence has to exist in the subject. There cannot be an
instance of sound that is not produced. The meaning of “flawless subject in question” is that the
subject has to be something we have a question about—in this case, whether it is impermanent.
There can be no doubt as to whether it is produced. In this case, we know all sounds are
produced, we assert that therefore they are impermanent, and thus we have determined that the
first mode, the subject quality, is present in the statement.
The definition of the forward pervasion is “a reason that has been determined to be
present only in the concordant aspect or aspects.”(8) The concordant aspect is that which is similar
in nature to the predicate. This means that the reason has to accord with the predicate, or in this
case, “Produced” has to be in accord with “impermanent. Furthermore, the reason cannot accord
with that which is dissimilar to the predicate, for example, “permanent.” which it does not, so
the forward pervasion is indeed present.
The definition of reverse pervasion is “ a reason that has been determined to be absent in
all instances of the non-concordant class.”(9) The non-concordant class is that which is
contradictory to the predicate. Non-concordant classes are divided into three types: those of non-
existent phenomena, those that are other than the concordant class, and those that are
contradictory to the concordant class. A traditional example of a non-existent phenomenon is
“the horns of a rabbit.” The non-concordant class of phenomena other than the concordant class
would, in this case, be things that are simply other than impermanent, such as “objects of
comprehension,” the premise of which does not directly relate to permanence/impermanence.
The non-concordant class of phenomena that are contradictory to the predicate would be,
in this case, the class of permanent phenomena. To the Vaibāṣhikas and by extension the
Sautrantikas, with whom Dignāga is associated, permanent phenomena include space, non-analytical
cessation or absence, and analytical cessation or nirvana. Thus, if a reason is found to
be present in any one of these non-concordant classes, the reverse pervasion does not occur, and
the reasoning is faulty. The lack of any one of these modes in any syllogism automatically
invalidates it even though it may seem true on the surface. The three modes provides a thinker
with a systematic way to evaluate a truth statement, whether it comes from an opponent in
debate, or from one’s own conceptual mind. It gives one a method to step outside one’s
preconceptions and rigorously examine the coherence of a statement multilaterally, uncovering
any internal inconsistencies.
Another important theory advanced by Dignāga is that of apoha, or exclusion, which is
basically the idea that the meaning of a term, as it is apprehended by conceptual mind, is actually
everything other than what the term is applied to. This is described as a kind of nominalism by
Chakrabarti and Siderits:
Buddhist logicians have an error theory about universals and permanent substances thatThis theory about how conceptual mind cognizes its objects, generally characterized phenomena,
they reduce to mental or physical particulars or simply eliminate. There are nothing but
momentary quality particulars in the world. But the human mind, afflicted by
perpetuation wishes and language-generated, deeply ingrained myths, has a tendency to
cluster some of them together first in the fictional form of enduring substantial
things...and then further classify these “things” into types...[w]hen a particular cow...is seen to be other than all other animals, the original indeterminate (concept-free)
perceptual content somehow causally triggers this difference-obliterating tendency. The
particular cow image is made to “fit” this linguistic and imaginative exclusion from the
complementary class of horses, rabbits, pillars, and such things. The specificity of the
particular cow—its its numerical detailed differences from other cows—is ignored;
instead, this mere exclusion from non-cows is foisted onto the perceptual content as a
predicate. This exclusion masquerades as the universal cowness.(10)
says that our assertions are often wrongly predicated due to mixing the particular with the term
generality. For Dignāga, the categories we create for things have no reality of their own, and
neither do they have any real connection to the specifically characterized phenomena they
represent. Even though the theory of apoha was challenged, especially by the Jainas and the
Mīmāmṣakas, It continued to be defended and adapted by Buddhist thinkers like Dharmakīrti
Refuting Other Schools
One practical purpose for the development of such definitions and theories by Dignāga
was to refute the views of opposing schools in debate. Of the six chapters of the
Pramāṇasamuccaya, five are dedicated to criticizing other schools of thought. He even refutes
his own teacher, Vasubandhu, though he does so in a roundabout way, by disputing Vasubandhu’s
authorship of a text called Vādavidhi.(12) Dignāga takes a definition in the Vādavidhi, that“perception is a cognition produced by that object,” and refutes it, saying
[If that which forms a cause of cognition, although it assumes an appearance different from its real form, is to be recognized as the object, then] there would also be the absurd conclusion that even the visual sense and the other [senses] would be [admitted as] objects [of cognition]. This is because they also exist, in the ultimate sense, in different forms [from those appearing in cognition], and [yet they] become the cause of such cognitions as the representations of something blue,etc., or of a double moon, etc.(13)In other words, perception is not just caused by its object alone, but by contributing factors such as the sense consciousnesses, and to ignore that would lead to absurd implications. Dignāga concludes that since the Vādavidhi is thus flawed, it must not have been composed by Vasubandhu, or if it was, it is a case of him being mistaken since he contradicts himself elsewhere.
One Hindu school of thought Dignāga takes on is the Naiyāyika. He begins with their
definition of perception: “That cognition which is produced through the contact of sense and
object, which is inexpressible (avyapadeśya), nonerroneous (avyabhicārin), and of a determinate
nature (vyavasāyâtmaka),”(14) and shows that the qualifiers in the definition are inadequate. The
sense consciousnesses, for example, could never perceive something expressible, since what is
expressible is by definition the province of conceptual mind. Therefore it is redundant to
describe the cognition of sense perception as inexpressible. Dignāga also takes issue with the
Naiyāyika view that physical senses actually go out to meet their objects outside the body.
Saying that, instead, the sense faculties remain in their bases in the body and perceive their
objects there, which, he says, is why we can see objects at a distance and why we can’t see when
our eyes are covered.(15)
Another school Dignāga criticizes is the Sāṁkya, who held that the five senses, through
the auspices of the mind, act to apprehend objects of sound, color, and so forth, that are equally
composed of the three guṇas, or potentialities inherent in everything. Dignāga responds that this
implies that (a) the sense organs would be infinite in number, because every object is just a
different proportional mixture of the three guṇas, and that means there are infinite possible
variations, each of which would need its own sense faculty; or (b) one sense organ would
apprehend all objects, because again, the objects are all of the same substance: the three guṇas.16
The whole theory of the three guṇas is attacked as being inapprehensible, either taken separately
or as a unity. He accuses the Sāṁkya of an “insufficiency of definition”17 in naming the type of
cognition that can apprehend mental objects, such as recollections, or that can recognize such
things as pleasure or desire. He also faults the Sāṁkya for asserting, by implication, that the
sense perception and mental consciousness operate simultaneously. Dignāga held the Sautrantika
view that a moment of perception was followed by a second moment of both perception and self awareness, followed by a third moment in which consciousness arose.
The final chapter of the Pramāṇasamuccaya examines the Mīmāṃsaka theory of
perception, in which they define perception as a cognition that arises “[w]hen a man’s senses are
in contact with something existent.”18 Dignāga takes exception to the use of the qualifier
“existent” (sat), pointing out that the use of the term “contact” naturally excludes the nonexistent,
since a sense faculty can only contact existents. This flaw renders the definition
essentially meaningless. Dignāga anticipates Mīmāṃsaka responses that the use of the term
simply denotes the counterparts, or objects of the senses by saying that the names of the actual
counterparts, i.e., sound, form, and so on, should instead be used as the predicate of the
definition. When his hypothetical Mīmāṃsaka interlocutor replies that the term “contact”
implies a transitive action toward objects, or even the mind and soul (ātman), Dignāga counters
that it has already “been proved that the soul...and other factors[i.e., the mind and the senses]
come in contact only with ‘sat,’”19 or existent phenomena. Interestingly, Dignāga does not
directly refute the existence of a soul at this point, which presumably would have been his position, but merely claims that a soul would not be able to contact non-existent objects.
Dignāga also counters the Mīmāṃsaka assertion that “[p]erception is that by means of
which an ascertainment...in the form of ‘this is a cow’ or ‘this is a horse’ arises in regard to ‘this’
[immediately perceived object].” This, he points out, amounts to saying that “sense-cognition is
able to perceive cow-ness” at the same time it perceives the particular cow it is looking at. For
Dignāga this is an untenable position because bare perception is incapable of perceiving “a
cow,” as such.
Therefore, in all cases of [our cognizing] a qualifier (viśeṣaṇa) with a qualified (viśeṣya)Again, Dignāga is reiterating the fact, which he considers settled, that the properties we ascribe
or a name (abhidhāna) with an object named (abhidheya), there is [involved] a
conceptual construction (vikalpa) produced by the mind (manas) which ascribes identity
(abhedôpacāra) [to the two factors], and [there is] not sense-cognition.(20)
to a category like “cows” are not perceived by the senses, but overlaid by conceptual mind.
Critiques of Dignāga
Some of Dignāga’s positions have been seriously questioned by other philosophical
schools, including the Mīmāṃsaka and the Madhyamaka. The question of how to interpret
Dignāga is still the subject of scholarly controversy today. In his book Buddhists, Brahmins and
Belief, Dan Arnold argues for an analysis of Dignāga as a foundationalist, which means that his
views implied previous beliefs as a valid basis for forming new beliefs. Arnold characterizes
Dignāga’s preference for direct perception as making either “...the ontological claim that mental
events are all that really exist or the strictly epistemological claim that mental events (such as
representational ‘sense-data’) are all we can directly know.”(21) He criticizes both Dignāga and
Dharmakīrti for the causal nature of their account of the relationship between “propositional
judgments” and “foundational perceptions,” and questions how experience could be the most
privileged source of knowledge when experience and perception seem to be completely divided
along the perceptual/conceptual line. Arnold paraphrases John Mcdowell in asking whether,
since knowledge, by definition, consists entirely of conceptual content (i.e., terms and
generalities), it can never really occupy the same “logical space”(22) as perceptual contact. “[I]n
the end,” says Arnold,
...the views of Dignāga and Dharmakīrti...involve an epistemic notion of truth. And suchArnold’s criticism is concerned more broadly with “epistemic notions of truth,” wherein the fact
a notion is at odds with what we can characterize as a realist notion of truth. Given these
points, Buddhist thinkers in the tradition of Dignaga and Dharmakirti are in a position
only to give reasons for why they believe what they believe, not for the likely truth of
of beliefs “being known” is held as evidence for them also being warranted.
Arnold also brings in Mīmāṃsaka and Madhyamaka critiques to bolster his charge of
foundationalism against Dignāga, as well as those later in his tradition. The position that the
Mīmāṃsaka take exception to is the Buddhist’s privileging of direct cognition, or perception, as
what is finally problematic about the Buddhist approach (and what renders thatThe result of this privileging leads, according to Arnold, to a kind of “epistemic circularity.”(25)
vulnerable to the Mīmāṃsaka critique) is simply the claim that one can specify that only
certain kinds of cognition have this status. What the Mīmāṃsaka have shown to be
problematic, in other words, is simply the view that one can specify that certain kinds are
uniquely able to confer justification, simply in virtue of their being of that kind—with the
question of which kinds are singled out ultimately being less significant...Wedded as they
are to an essentially causal approach to justification, the Buddhist critics of the
Mīmāṃsaka doctrine of intrinsic validity do not appreciate that their attempt to privilege
causally explicable cognitions only presupposes precisely what is called into question by
What the Mīmāṃsaka have in common with the Madhyamaka, says Arnold, is the fact
that they both require an epistemology that is in accord with intuitive experience—counter to the
Sautrāntika redescription of reality in terms of relative and absolute truth in which actual reality
is said to be quite different from what we experience. The Madhyamaka took the ontology of the
Abhidharma, rather than the epistemology of Dignāga and Dharmakīrti, to be the salient
contribution of foundational Buddhism—specifically, the doctrine of interdependent origination
(pratītyasamutpāda). The Madhyamaka considered that the
ontological primitives posited by Abhidharma could have explanatory value only if theyThe Madhyamaka, according to Arnold, rejected the epistemological project of Dignāga largely
are posited as an exception to the rule that everything is dependently originated; that is,
dependently originated existents would really be explained only by something that did
not itself require the same kind of explanation. But it is precisely the Madhyamaka point
to emphasize that there is no exception to this rule; phenomena are dependently
originated all the way down, and it is therefore impossible to specify precisely what it is
upon which anything finally depends. Hence, there is no set of ‘ultimatly existent’
on the grounds that there are no truly existent objects for a valid cognition to perceive, and that
any knowledge claims are therefore flawed and that such claims involve infinite regress. For the
Madhyamakas, there simply is no basis for our conventions.
Christian Coseru criticizes Arnold’s analysis for being predicated on the view that the
idea of self-aware valid cognition necessitates the conclusion that for Dignāga, “mental events
are all we can directly know.”(27) Coseru makes a case for interpreting Dignāga “as advancing
something like a phenomenological account of cognition.”(28) In this account, the resemblance of
subject and object is accounted for by the category of self-aware valid cognition, without which
“object-cognition without self-cognition and self-cognition without object-cognition would be
indistinguishable.”(29) Coseru criticizes Arnold’s Kantian view that all perception is self-conscious
and argues that the “...justification of knowledge—which is the primar[y] concern of
foundationalism—falls outside the scope of [the] pramāṇavāda enterprise,”and offers “an antifoundationalist reading of the Buddhist epistemological project” that does not rely on “...the
requirement that perceptual awareness provide a justification for basic empirical beliefs.”30
Coseru dismisses Arnold’s concerns with the perceived inherent contradictions of self- aware
valid cognition, saying “...self-awareness is precisely the type of cognition in which there is
awareness that (or thinking) is taking place without there being any engagement in it. Such is the
case, also, with feelings and memories.”(31) Coseru questions whether “anti-psychologism is still
tenable as a philisophical attitude” in light of advances in cognitive science and a return to
naturalism in epstemology, and points out that the primary purpose of Buddhist epistemology is
The Path of Conceptual Mind
Dignāga's teaching, then, should be valued and studied mainly for it’s value as a path
toward the Buddhist goal of liberation. Its foundational relationship to the scholastic traditions of
Tibetan Buddhism is undeniable, and those traditions continue to give practical application to
Dignāga's theories in the form of debate, or “clear thinking” practice. The point of debate is to
train the mind, through repeatedly challenging assertions and having one’s own assertions
challenged, to question one’s own views rigorously. The ways that suffering is created are
conceptual—it is considered that saṃsāra, or cyclic existence permeated by suffering, results
from kleṣa, or afflictive emotions. Kleṣa, in turn, result from conceptual mind. Thus, the root of
suffering lies in conceptual mind.
That does not mean, however, that conceptual mind should be discarded or ignored (as if
such were possible), but that conceptual mind is the very way to examine and dismantle the very
beliefs that give rise to the narrative triggering of afflicted states of mind. As Nāgārjuna is
supposed to have said, conceptual mind appears at the beginning to be one’s worst enemy; later
to be one’s spiritual friend; and finally, to be the Buddha. This describes the contour of Buddhist
soteriology—conceptual mind is the only kind of consciousness that can grasp logical argument
and identify sound reasoning, which leads to inferential valid cognition. In the end, when
realization is gained, inferential valid cognition is transformed into yogic valid cognition, which
sees reality as-it-is and in its entirety. This means, paradoxically, that although conceptual mind
is the problem in the first place, there is no way to reach liberation other than through conceptual
One can meditate intensively for many years, but if conceptual mind is not engaged, it is
unlikely to lead to liberation—even if it does, conceptual mind has to come along eventually.
That is why the schools of Tibetan Buddhism have all made the practice of debate and the study
of logic part of the formal education of monks—without clear thinking, the training of these
monks to be bodhisattvas would be incomplete. While meditation and devotional practices might
increase compassion and commitment, only the practice of logic can train conceptual mind.
Conceptual mind is always considered to be mistaken, in that it does not deal with
existents, but with mental constructs, and only apprehends its object indirectly. However, it is not
always wrong. The output of conceptual mind is always in the form of truth statements using
terms and categories. Through the application of the three modes to the truth statements we
make to ourselves, we can root out the beliefs those statements are based on which are not
themselves based on valid cognition. Thus arriving at correct inference—which, though still mistaken, is valid, and it can lead to a certain kind of conceptual realization. In other words, it
can lead to a mystical experience. Anne C. Klein says the connection between the direct and the
conceptual is a natural one:
The intellect’s significance for mystical experience would also be impossible except forThis would seem to answer some of Arnold’s concerns with what he sees as a chasm between
the assertion that the nature of awareness itself is such that conceptuality can be
transformed into direct perception...conceptual thoughts and direct perception are not
merely compatible in terms of their potential for mutual enhancement, but their very
nature is the same.(32)
perception and inference. According to Klein, since the definition of mind from the bLo Rigs (33) is
“that which is clear and aware,” and direct perception and inference are both minds, they are of
the same essence, and hence there is no real contradiction, but they are more like different
modes. It also points again to the how the ideas and definitions compiled by Dignāga are used as
a path—epistemology in service of soteriology. In describing the scholastic tradition of the
Gelukpa sect of Tibetan Buddhism, Klein lays out the contour of this path:
Conceptual thought is not transformed of its own force into direct experience; otherIn Buddhism, wisdom and compassion are said to be like the two wings of a bird—without both
elements of mental training and engagement with religious path are involved. Most
significant is the gradual union of conceptual understanding with increasingly stabilized
periods of concentration. The practitioner alternates between analytical meditation
(dpyad sgom) and stabilizing meditation (‘jog sgom). Finally, one reaches the point where
instead of analysis acting as an interference to stabilization, or stabilization weakening
analytical understanding, each enhances the other. This union of analytical or insight
practice with a concentration developed to the point of calm abiding (śamatha, zhi gnas)
is called special insight (vipaśyanā, lhag mthong). Attainment of this union of insight and
concentration marks the beginning of the second of the five paths or stages of practice,
the path of preparation (prayoga-mārga, sbyor lam).(34)
the bird can not fly. Dignāga can be credited with codifying the technicalities of the path of
conceptual mind, and for this he is rightly esteemed. His is not seen as as the final word, but
understanding and practicing his way of thinking is indispensable in causing the development of
1 Hattori, 1.
2 Ibid., 6-10.
3 Ibid., 3.
4 Arnold, 3.
5 Coseru, 241.
6 Tenpa, coloected topics.
7 Acharya Sherab Gyaltsen, commentary, 5.
8 Ibid., 12.
9 Ibid., 22.
10 Siderits, 12.
11 McCrea, Patil, 20,21.
12 Hattori, 114.
13 Hattori, 35. Brackets in original.
14 Ibid., 36.
15 Hattori, 38.
16 Ibid., 52.
17 Ibid., 61.
18 Ibid., 63.
19 Ibid., 64.
20 Ibid., 67.
21 Arnold, 35.
22 Ibid., 41.
23 Ibid., 48.
24 Ibid., 102.
25 Ibid., 103.
26 Ibid., 119.
27 Coseru, 243.
28 Ibid., 244.
32 Klein, 16.
33 The Classification of Minds, the Tibetan text that assembles the definitions and categories of Dignāga and
34 Klein, 16.
Arnold, Dan. Buddhists, Brahmins and Belief. New York: Columbia University Press, 2005.
Coseru, Christian. “Naturalism and Intentionality: A Buddhist Epistemological Approach.”
Asian Philosophy 19, no. 3, November 2009, 239–264.
Hattori, Masaaki. Dignāga, On Perception, being the Pratyakṣapariccheda of Dignāga’s
Pramāṇasamuccaya. Cambridge, Harvard University Press, 1968.
Klein, Anne C. Knowledge and Liberation: Tibetan Buddhist Epistemology in Support of
Transformative Religious Experience. Ithica: Snow Lion, 1998.
McCrea, Lawrence J., and Paramil G. Patil. Buddhist Philosophy of Language in India:
Jñānaśrīmitra on Exclusion. New York: Columbia University Press, 2010