BUDDHIST

EPISTEMOLOGY
Recent Titles in
Contributions in Philosophy
The Last Choice: Preemptive Suicide in Advanced Age, Second Edition
C. C. Prdo
Self-Construction and the Formation of Human Values: Truth, Language, and Desire
Teodros Kiros
Diogenes of Sinope: The Man in the Tub
Luis E. Navia
Simone de Beauvoir Writing the Self: Philosophy Becomes Autobiography
la-Ann Pilardi
Stalking Nietzsche
Raymond Angelo Belliotti
The Transient and the Absolute: An Interpretation of the Human Condition and of Human
Endeavor
Mordecai Roshwald
The Adventure of Philosophy
Luis E. Navia
Intentionalist Interpretation: A Philosophical Explanation and Defense
William Irwin
Natural Law Ethics
Philip E Devine
Anglo-American Idealism, 1865-1927
W l. Mander editor
Two Views of Virtue: Absolute Relativism and Relative Absolutism
F F Centore
Liberty: It" Meaning and Scope
Mordecai Roshwald
BUDDHIST
EPISTEMOLOGY
S. R. Bhatt and Anu Mehrotra
Foreword by the Dalai Lama
Contributions in Philosophy, Number 75
Fran J. Hofman, Seres Adviser

GREENWOOD PRESS
Westport, Connecticut London
Libra of Congres Cataloging-in-Pbliction Data
Bhatt, S. R. (Siddheswar Raeshwar), 1939-
Buddhist epistemology I S. R. Bhatt andAnu Mehrotra ; foreword by the Dalai Lama.
p. cm.-{Contibutions in philosophy, ISSN 0084926X ; no. 75)
Includes the text of Dig mig a's Nyayamukha.
Includes bibliogrhical references and index.
ISBN 0-313-31087- (a paper)
1. Knowledge, Theor of (Buddhism) 2. Dignaga, 5th cent. Nyayaukha.
3. Buddhist logc. 1. Dignaga, 5th cent. Nyayamukha. II. Tite. II. Seres.
BQ44 0.B53 2000 99- 513
British Librar Cataoguing in Publication Data is available.
Copyright 2000 by S. R. Bhatt andAnu Mehrota
All rights resered. No portion of this book may be
reproduced, by any process or technique, without t
h
e
express written consent of the publisher.
Libr of Congrss Cataog Card Number: 99-04513
ISBN: 0313-31087-4
ISSN: 0084-926X
First published i 2000
Greenwood Press, 88 Post Road West, Westport, CT 06881
A imprint of Greenwood Publishing Group, Inc.
www.grenwood.com
Prnted in the United States of Amerca
The paper used in ts book complies with the
Peranent Paper Stadad issued by the National
Inforaton Standards Oranization (Z39.48-1984).
10 9 8 7 6 5 4 3 2 1
Contents
Foreword by the Dalai Lama
Preface
Introduction
1. The Buddhist Theory of Knowledge
2. The Buddhist Theory of Perception
3. The Buddhist Theory of Inference
Appendix 1: NaaprveSakasatrm
Appendix 2: Naaprveiakaszm (Translation)
Not es
Glossary
Bibliography
Index
vi
i
11
25
49
101
105
111
125
129
135
Foreword
Te Dalai Lma
When Buddha
S
akyamuni attained enlightenment beneat the Bodhi tree more
than two and a half thousand years ago, his achievement ws not only the result
of having reached the peak of meditative stabilisation, of having brought great
compassion to fuition, but also of clear analytic tought. The lucid simplicity of
his subsequent teachings are ample evidence of this. And indeed he encouraged
his followers to regard even his own advice in the same rigorously critical light .
Thus, the study of logic and te nature of kowledge have been crcial to Buddhist
tradition from the outset.
These disciplines contiued to develop and fourish i India afer the Buddha' s
passing away, reachig their acme in the works of Vasubandhu, Dinnaga and
Dharmaklrti, which were to become seminal to stdies in the great monastic
universities, such as Nalanda and Vikamasila. And it was this tradition of acute
logic and analysis tat was transmitted and preserved in Tibet for more than a
thousand years, where it was employed not merely to challenge the views of
others but to ensure te clarity and authenticity of on
e
' s own view.
Therefore, I an1 delighted tat two distinguished contemporary Indian scholars,
Dr. S.R. Bhatt and Dr. A. Mehrotra, have written the present book on Buddhist
Epistemology, including an English translation of te "Nyaya Pravda. " This
valuable work sheds light on abstruse topics and will allow readers to gain a
clearer appreciation of the depts of Buddhist knowledge.
Preface
The present work is an analytical exposition of the teory of knowledge as
propounded in te DiIaga-DharakIti tradition. It expounds te Buddhist theory
of kowledge in its totality.
Buddhist tinkers unaniously uphold the view that tere are two kids of
objects of kowledge, namely, the unique particular (svalakma) and the
generalized image (stmtnyalakalJa) . Unique particular is discrete and
istantaneous being. I is unique particular in the sense that it is neiter identical
nor siilar to other unique particulars. It is te only objective real (armtnha
sat) . As distinct fom unique particular there are objects tat are constructs of
our intellect and that are in te form of generalized iages . They are subjective
i origin but intersubjective in nature.
On te basis of te precedig analysis Buddhist tinkers emphatically maintain
that sice there are only two kids of objects of kowldege, there are only two
kinds of kowledge (rmta) namely, perception (rtaka) and inference
(anul1ltna) . The unique particular is amenable to perception only, whereas
generalized image is kown through inference only. By implication, unique
paricular can never be kown trough inference and generalized image can
never be kown trough perception. Thus, each of te two types of kowledge
has its own separate and distinct sphere of operation.
In the followig pages is a discussion of Buddhist teory of kowledge,
comparig it with oter schools wherever necessary.
To make the logical issues more intelligible to readers, a basic text, " Nyaya
Pravda" is given as an appendix of the work. The text of " Nyaya Praesa" is
generally ascribed to DiInaga, a pioneer thinker in Buddhist teory of Kowledge.
DiInaga provided a solid footing to Buddhist epistemology and logic and gave it
a distinctive character. He has been regarded as te fther of Buddhist epistemology
x Preface
and logic in particular and of the entire medieval Indian epistemology and logic
i general . There are several works composed by him i this area, among which
"PramaIa Samuccaya" and "Nyaya Pravda" are most iportant . " Pramala
Samuccaya" is not available in its complete foml, even though some of its chapters
have been restored fom Tibetan and other sources. However " Nyaya Pravesa" is
available to us in its fll form. An attempt has been made here to translate it in
English for te beneft of English language readers. The text is presented here in
Roman script with a view to have wider readership. Explanatory notes have also
been provided to clarif some kotty points.
" Nyaya Pravesa" is a classical work pertaining to the modes of kowing and
reasoning in the Buddhist tradition. For centries it has been studied as a manual
of the Buddhist theory of kowledge in India, Tibet, Chia, and oter countries.
It provides a foundation to the famous Buddhist art of debate (vtda-vidhi) . There
are several Tibetan and Chinese commentaries and sub-comentaries on " Nyaya
Pravda" apart fom the ones in Sanskrit. This accounts for its signifcance and
popularity among classical scholars.
In a cryptic, but succinct, manner "Nyaya Pravesa" presents Diaga' s views
on the nature of perception and inference and their fallacies . It also elaborately
discusses the modes of argumentation and reftation along with teir fallacies.
We are grateful to Greenwood Publishing Group and its staf and to
Dr. Hofan who introduced this work to Greenwood. Our thans are also due to
Indian Council of Philosophical Research for providing fnancial assistance for
the preparation of the work.
Itroduction
DICHOTM OF SUSTANCE AND NO-SUSTANCE ONTLOGIS
The varied and multifaceted Indian philosophical thought is characteried by two
broad philosophical tendencies tat are antitetical in natre but both of which
can be traced in germinal form to te Upaniadic tought. One tendency, which
can be termed " CtavCda" or "substance ontology, " has been dominantly
presented in te so-called astika systems, te culmination of which is found in
the Advaita Vedanta . The other tendency, which is popularly known as
" anatavCda" or "no-substance ontology, " fnds its advocac and manifestation
i Buddhist tought. The basic contention bifrcating the two tendencies is te
view regarding the ontological status of permanence and change, bot of which,
though opposite i nature, are given to us i veridical experience, and therefore
bot claim the stats of reality. The substance ontology advocates the permanence
aspect and tries to explain the phenomenon of change. The no-substance ontology,
on the contrary, accepts the reality of change alone and explains the experience
of permanence as conceptual superimposition. Thus, the history of Indian
philosophical thought could be approached in terms of a dialectic of these two
divergent tendencies.
The substance-ontology revolves around te idea of permanence or abidance
as te sole criterion of realit. Accordingly, substace, which alone is te substratm
of all attributes and modes, has the sole reality or te primary reality, and te
attributes and modes have either an apparent existence or a derivative existence.
Accordig to Advaita Vedanta substance, which is a unitary, homogeneous, pure
consciousness, alone is real, and all attributes and modes are phenomenal in the
sense that they are mitha (i . e., neither real nor ueal). The Sarkhya system
and also the Nyaya-Vaiseika and MIamsa systems, however, assign some reality
to change also insofar as matter is regarded as subject to mutation wheter in te
2 Intoducon
form of evolution from one mass of matter or in the form of a combination of
different elements of matter, thereby producing a di f ferent or new
complex. But the point to be noted is that even in these schools that
display a realistic tendency the primacy of substance and its permanent
essence are emphasized. It may be signifcant to poin out that there has also
been another tendency in the school of Jainsm wherein the exlusiveness of
substance ontology and no-substance onology hs been rejected, and equal status
is accorded to both permanence and change.
NO-SUSTACE ONOLOGY OF BUDDIDSM
Buddhist thOUght has centered around no-substance ontology. Gautama,
the Buddha, who initiated Buddhst thought, was led to philosophizing by a
intense longing for the eradication of sufering. He visualized that all sufering is
due to 'tauM' (longing) and that all longing is de to attacent to the false
notion of pennanence. He wnted to suggest a way out of the labyrinth of suffering
and put forth impennanence or nonsubstantivality as the key to overcoming
suffering. Te four Noble Trths, therefore, advocate the idea of impenanence
and nonsubstantivality as the way to eradicate suffering. The entire reality that is
generally understood in tenus of matter and conciousness has been understood
by the Buddha as a series of changing moments (kalw). The word "moment" is
suggestive of the fact that all real or existent is time-embedded. All that exsts,
exists in time. Therefore, an existence series is identical with time series. The
existence series could be physical or psychical or a conglomeration (sarighata) of
the two. Every series in itself is also a conglomeration. Material entity is a
physical conglomeration (bMt sarigMta), whereas mental entity is a pschical
conglomeration (citta sarightta). A living being is a psychophysical
conglomeration. Since the psychical conglomeration is of four types, a
psychophysical conglomeration is named as pafcaskndha, consisting of four
psychical and one physical conglomerations. Tle fupsical coerton a
vi s, smiskr skrwha, nn skndha, and sarlia skrwha. The
only physical conglomeration is rapa skandha.
The existence series, whether of a physical conglomeration or pschcal
conglomeration, consists of distinct unts of existence that are discrete but not
independent. They are rather interdependent in the sense that every succeeding
unt in the series .ls causally dependent on its preceding unt in that series, and
liewise every preceding unit in the series, in order to be real, has to give rise to
its succeeding unit. These two ideas are tecnically known as prtita samutpada
(dependent origintion) and ara kyakarta (causl effcienc). Tley consttute
the essence of the four Noble Trths propounded by the Buddha ..
The concept of existence series (ka{la santana) is pivotal to Buddhst
metaphysics. It explains the reality of incessant change in tenns of the causal and
dependent origin of the succeeding from the preceding unit in the series. The
experience ofpennanence in the series is also explained on the basis of continuity
and dependence. The real is always changing, but the change is not random; it is
Intoduction 3
causally regulated. In case of a human being, te series is psychophysical, having
a distinct idyntity of its own. But this poses the problem of memory, recollection,
and recognition in every distinct personal identity. Buddhist thought explains
these problems also in terms of karza santana (existence series). In every
psychophysical series that is unique and self-identical te succeeding stems fom
the preceding, and this conglomeration of fivefold series, which begins with the
birth of the conglomeration in a particular form, continues till the death of that
conglomeration i that fon. From birth to death in this series there is a constat
change, and yet there is retention wit the possibilit of recollection and recognition
on the basis of which te past is retained in the present and handed over to the
fture. In death the total conglomeration does not cease to exist or does not come
to an end and gets retained so as to give rise to anoter conglomeration in some
other form in te next birth. In this metanlOrphosis the psychical series is present
only in the fonn of samskars (latent ipressions) sustained by karic forces.
The kannic forces determine the nature and fon of the next birt. They also
determine the pafcaskandhas (the fvefold conglomeration), which has to come
into existence in te next birth. There is continuity not only witin one particular
birth but also fom one birth to another birth. Thus, i Buddhist thought we
fd a remarkable explanation of the problems of personal identity, te experience
of permanence, and so on expressed in the phenomena of memory, recollection,
and recognition.
There are some interesting implications of the nocsubstance ontology
advocated by the Buddha. The idea of impemlanence of realit results in the
denial of a pemlanent and in1lOrtal soul. However, this does not mean that there
is no eschatology or soteriology in Buddhist thought. The Buddhist account of
the destiny of the individual self is conditioned by the ideas of anitata
(ipennanence), pafcaskandha (fvefold conglomeration), and santana( existence
series), and accordingly, the destin of the individual self is characterized as
attainment of nira1a.
Another implication is i the form of insistence on human autonomy and
denial of any supraluman or divine agenc for helping the human being to attain
nira1a. Buddha spearheaded the srma1a tradition, which emphasized self
effort and noble conduct (ara-a!a1gika marga) . In this scheme tere is no role
assigned to any divine or superhuman agenc. Buddha always inlpressed upon his
followers to have a critical attitude and an analytic bent of mind. He was opposed
to blind fith, and that is why he rejected all notions about transcendental entities.
However, it does not mean that Buddha was antispiritalist. He had an abiding
fait in a moral and spiritual order. That is why he was not a materialist like a
Caraka.
mSIRY OF BUDDHIST THOUGHT
Early Buddhism centers around the teachings of the Buddha, which are
contained in the Pali Tripiakas. This phase of Buddhist thought is therefore
4 Introuction
popularly kown as Pali Buddhism. Buddha was not a mere speculative and
individualistic thinker. He acquired knowledge and enlightenent not for the
eradication of his own sufering but for the eradication of suferig of all living
beings. Thus, his thought has a practical and universalistic dimension that is
essentially correlated with a teoretical aspect. The teachings of the Buddha
were for the beneft of all living beings who have the capacity to acquire knowledge
and to practice the noble conduct professed by him. In his thinking he did not
concentrate on iticate philosophical issues, this task was subsequently perfored
by his followers. The history of Buddhist thought evinces two tpes of following
of the Buddha, one philosophical and the other theological. Theologically, his
followers form two camps known as Hlnayana and Mahayana, each having
many sects and subsects. Philosophcally, Buddht tought i India h four brad
divisions, namely, Vaibhaik, Sautrantika, Madhyaika, and Yogacara. NarayaIa
Bhatta in Manameodaya crptically summarizes the basic thrust of each of these
four schools as follows:
Mllkho Mtdlzamiko vivartamakilam .nasa mene jagat,
Yo gtctr mate til santi mataya/l ttstm vivarto' khilal
Artho'sri laJik tvastvanwnito ktlzeri Salltrtnriko.
Patalalll laJa bhaJgllrlll ca sakalall Vaibhik bhtate.
Tat is, te Madhyamia is te most inlportnt philosophica school, which regards
te entire world to be an apparent manifestation of sana. The next school i s
Yogacara, according t o which ideas alone are real, and te entire world i s an
apparent manifestation of ideas. The tird school is that of the Sautrantikas,
which maintains that there are objects existing independently of the ideas, but
they are momentary, ad teir existence is known ony trough inference. The
last school is known as Vaibhaika, which holds that all objects are momentary
and perceivable.
For the sake of eas uderstandig, tese fou schools can be pu i a particuar
logical order as Vaibhaika, Sautratika, Madhyanlika, and Yogacara. Perhaps
tllis may also be te chronological order. All these four schools claim authenticity
and faithflness to Buddha's tought. Buddha was quite prolifc in his ideas, and
it may not be di fficult to find rudiments of all these four schools in his thoughts.
The doctrine of impermanence (anitata) , which subsequently was formulated as
te doctrine of momentariness (
k
mikavada) , and the doctrine of dependent
origination (rtita samutada) are te conmlon pla for all tese four schools.
All the schools believe in te ideas of rebirth and nirvala. Their diference is
ony in terms of the development of logical structures based on these doctrines
and ideas. We shall briefy discuss the major points of di fference characterizing
each of tese schools. The schools of Madhyanlika and Yogacara (Vijfanavada)
are idealistic in teir thrust ad are associated with the Maayana tradition. The
other two schools have a realistic tendenc and are put under Hlnayana tradition.
However, tese four schools constitte signifcant facets of later Buddhist thought.
Introduction
5
VAIHIA SCHOOL
The Vaibhaika school seems to have acquired this name because it relies
upon a co
m
mentary on the Tripi/aka kown as Vibla. This school upholds a
dualism of mental and physical elements. All.tat is real is momentary, and it is
either physical (bMt) or mental (cita) . The objects of the world are a
conglomeration of either physical elements or mental elements or both. No object
is permanent or abiding, but sice it is i te form of an incessant series, it gives
the impression of permanence. These objects are directly kown in perceptual
cognition. Wen there is sinlUltaneit of appearance of te cognzig consciousness
and the cognized object at a particular time, perceptual cognition of that object
takes place. Thus, all three have simultaneous origin at one and the same tie.
Though te object depends on the cognizing consciousness in order to be kown,
the dependence is not necessary for mere existence. A object may exist without
being kown and may thus be independent of the cognizing consciousness.
Liewise, a cognizing consciousness may also exist independently of the cognied
object. In this way, the Vaibhai school advocates dualism i its metaphysics
and realism in its epistemology. The Vaibhaika school concentrates more on the
analysis of mental phenomena, and we fid an elaborate psychological analysis
i te literatre belonging to this school.
SAUIA SCHOOL
The Sautrantika school derives its nanle from a commentary called Sitranta.
It is also a realistic school, sharing its metaphysics with Vaibhaikas. Sautrantikas
believe in the momentary existence of the real and classif them into mental and
physical. The mental and the physical are basically independent of each other,
though subsequently they may interact. Sautrantikas regard the object to be
existing independently of the noetic process. The object, the cognizer, and the
cognition are all distinct. The object may be known or may not be blown.

If an
object is known, it canot be known in direct perception. In this respect tey
differ from Vaibhaikas. According to Sautrantikas, the moment of existence of
tlle object and tlle moment of the cognition of the object canot be the same
because cognition follows and presupposes existence, and every existence precedes
its cognition. There camlot be simultaneity of existence and its cognition. The
Sautrantika tinkers examine and refte te Vaibhaika position in tis respect.
In fact, this is the major point of di fference between the two schools. The
Sautrantikas argue tat the object of knowledge exists independently of the
cognizing consciousness, and it ceases to exist in the next moment. So, when it
comes into existence at that very moment, it is not cognized, and it cannot be
cognized. I does not exist in te next moment to be perceptually cognized, so
there is never any perceptal cognition of an object. However, it does not mean
that an object can never be cognized. There is another mode of cognizing an
object. Before an object ceases to exist, it leaves out its ilpressior. This impression
is te exact copy of its original and has semblance (san/pya) wit it. The cognizing
consciousness perceptally apprehends only this impression, and through
6 Itoducon
tis perceptual apprehension of the impression it infers the original object. Thus,
Sautrantikas introduce the concept of object impression ad through this advocate
the representative theory of perception. For them all cognitions are represented
cognitions of the object. A object-qua-object is not directly perceived. Only its
impression is perceived, and because of te resemblance of the two and cognitive
nonavailabilit of the object te impression is taken to be the origial object.
MHA SCHOOL
The third school is Maa, which is philosophically very signifcant.
Nagarjuna has been the frst known exponent of tis school. He argu
e
s tat
Budda's teachings consist i te madamaprtipad (middle pat) to be followed
for te realization of nirvi1a, which is cessation of all suffering. According to
Nagarjuna, Buddha advocated sanati (essencelessness) of all existence. All
existences are sana (essenceless) i the sense tat tey do not have self-existence
(svabliva) . Every existence has a borrowed existence or a dependent existence
(prtltasamutpanna); svabhiva sanyati (lack of independent existence)
characterizes all reals. The same position holds good i respect of all tought and
laguage. Just as all real is self-negating, all tought and all language are also
self-negating. Nagarjuna exposes the hollowness and self-contradictory nature of
the inlportant concepts and doctries prevalent in his tinle in Buddhist and non
Buddhist philosophies. He successflly employs te weapons of slnati and
prtftyasamutida to demolish al systems of metaphysics. He advocates a twofold
approach to reality in terms of samvri sat (empirical real) and parmtrtla sat
(trancendental real) . Bot are characterized by sanuti in diferen ways. Empirical
is svabhiva sana (devoid of intrinsic existence) , and transcendental is prapaica
sanya. By prpaica he means display of thought and language. Nagarjuna
emphasizes te antinletaphysical, practical, and pragmatic nature of Buddha's
teachings and lays stress on the atainent of prjii (wisdom) leading to sUa
(noble cond
u
ct) and samidhi (state of equipoise) . This school is known as
Madhamika because of its emphasis on madfama prtipada, which is a practical
middle path, avoiding all extremes. It is k
n
own as
S
iyavada because of its
exposition of essencelessness of all real on accoun of its dependent character.
YOGACARSCHOOL
The pluralistic and realistic philosophy of the Sarvastivada culminates in a
monistic and idealistic philosophy of Vijfanavada (Yogacara) by way of a critique
and rejection of
S
unyavada. The representative theory of perception of the
Sautrantikas implies that all that is cognized is the content of cognition, and the
content of cognition has a form of its own, which has sameness of form (sirupa)
with the form of te object. It resulted in the teory of sikirjitnavida, a
teory accordig to which every cognition has a fom of its own apart fom the
form given to it by its object. From the theory of siktrjiinavida there was a
natural trfnsition to the theory that only contents of cognition or viiinas (ideas)
Introduction 7
are real and that they alone are cognized. The external objects are only
hypostatizations. They are presumed to be there, but, in fact, they are only
projections
o
f te consciousness. Consciousness alone is real. Thus, in Vijfaavada
we find a repudiation of te theory that te object of cognition exists exterally
and independently of the cognizing consci o
u
sness (blhyarthavada) . The
consciousness that alone is primarily real is momentary and is in the form of a
continuous fow (rvtha) . The Vijfanavada advocates tree levels of reality,
namely, parikalpita (iaginary), ptrtantrika (dependent or empirical realit) ,
and ptrmarthika (transcendental realit). Transcendental reality is conceived to
be unitary stream of consciousness, technically kown as ataa vifana. It is a
storehouse of consciousness in te sense tat it is the foundation of all streams of
consciousness that are responsible for te appearance of the world of external
objects. It is a repository of the old impressions (smiskars) and a depository of
new impressions. In this way alaa viftna is the only ultimate realit. In itself
it is pure and canot be characterized. It can be experienced ony in the state of
nirvtIa
.
The other level of realit is empirical, which consists of [mite stream of
consciousness and objects of consciousness. It is the level of the empirical world
that we experience in our ordinary life. It is a reality that is amenable to empirical
kowledge and linguistic expression. It is empirical realit (samv, sat) , as diferent
from transcendental reality (armtrthika sat) .
DEVELOPMN OF YOGACAR EPISTEMOLOGY
Maitreya and Asanga
In the Yogacara tradition prior to Maitreya,
A
sanga and Vasubandhu initiated
systematic philosophical refections. It seems Maitreya reached Yogacara position
via
S
uyavada, whereas Aanga and Vasubandhu were initially Sautrantik and
Vaibhaika, respectively, before coming to Yogacara. Maitreya ws the author of
several works such as Bodhisattvacart Nirdeta, Saptadasa Bhami
S
astr
Yogacara, and Abhisamaalafkar Karika. He discussed in detail the nature of
reality and the modes of kowing. In fact he is the forerunner of the art of debate
(vtda-vidhi) in the Buddhist tradition. Aanga followed Maitreya and expounded
Yogacara philosophy. He composed two work entitled Prkarltra Vacistr
and Mahaana Abhidhanna Sam;ukta Sangiti
S
astr. By and large he followed
Maitreya but differed in respect of te theory of proof (stdhana) .
Vasubandhu
Vasubandhu carried forward and systematized the Yogacara philosophy. He
was an autor of several inlportant works on ontology and epistemology. Some
of the works associated with hinl are Vada Vidhi, Vtda Marga, Vada Kallsaia,
Tarka
S
astr, and Abhidhara Ksa.
Diimaga
Among the post-Nagarjuna Buddhist philosophers DiIaga has been the
8 Introduction
most notable thinker who has carried ahead the tradition of Maitreya,
A
saIga,
and Vasubandhu by successflly repudiating te negativistc dialectics of Nagarjuna
and by replacing it with a philosophical positivism that was a synthesis of the
Sautrantika and Vijianavada schools. It goes to the credit of Diiaga that he
recolciled the Sautrantika ontology with the rubric of Vijianavada ontology,
formulated an epistemology and logic suited to tis new ontology, and propagated
a distinct philosophical methodology based on the technique of apoha (double
negation) . He introduces a new mode of philosophizing by iterspersing
ontological discussion in an epistemological settig, a stle that later on fascinated
te Nava Nyaya thinker GaIgesa and his followers. Diaga employs te technique
of double negation (apoha) for clarity and precision in thought and language. He
is an abstruse logician. In his logic he advocates a dichotomous classifcation.
This dichotomy is refected in the advocac of two tpes of real, two modes of
their cognition, a to shades of meanig. He applies the idea of svatovavartana
(mutal exclusion) in the felds of reality, knowledge, and language. In other
words, every real is a class in itself, and everything else constitutes i ts
complementary class. Thus, there is an exclusive ordering in the realm of reality
such that there are two and only to ty pes of real, namely, svalakala (unique
particular) and samanyalakala (its generalized image) . Whatever is unique
particular calot be its generalized image, and vice versa. The same rigid
dichotomy is extended to the feld of epistemology. It was i1ovative of Diiaga
to point out that every epistemology has to be structured keeping in view the
requirement of ontology. Following this ontological commitent in Diiaga's
epistemology, there are two and only two valid modes of knowing, namely,
prtaka (perception) and anumana (inference) , and only to ty pes of kowledge,
perceptal and inferential. The perceptual cognizes the unique particular alone
and can never cognize its generalized image. The iferential, on te other hand,
cognizes the generalized iage alone and calot cognize the unique particular.
So, just as there are rigid dichotomy and strict ordering at the level of reality,
there are also a rigid dichotomy and strict ordering at the level of kowledge.
Diiaga carries forward this dichotomous distinction to the field of language and
meag. Every concept is expressible in language i terms of its meaning. Meang
is conveyed by a word that represents a particular concept. The meaning of a
word is the negative of its negative. Thus, there calot be any overlapping or
cross-division i the meanings of any two words. Another great contribution of
Diiaga is in the form of propagation of the theories of:
1. Meddhlldmdnasiddhi(l: epistemology is to be constructed to support ontology.
2. PrmdlJa vavasthd: the separation of the two types of mode of knowing (rmdlas) .
3 . Hew caka : the theory that provides a schema of foral logic.
4. Anydpoha: the theory of double negation, applied to the fields of ontology,
epistemology, and theory of language.
The magnum opus of Dirnaga i s PramalJ a Samuccaya wi th i t s
Introduction 9
autocommentary (Svavrttc) , Naya Pravda, Naa Mukha,
A
lambana Parika,
and Hetu Cakr pamaru are some of his major works. Though PramaIJa
Samuccaya IS lost i its Sanskrit original, it has been partly restored from its two
Tibetan versions and Sanskrit fagments.
Dharmakri
Dharmaklrti not only mastered te systems of Vasubandhu and Dinaga but
also excelled in tem. Seven works of DharmakIrti are kown and available.
Among tem PaalJa Varika is most notable. Pamaza Varka is an advncement
of the view of DiIaga expounded in PrmalJa Samuccaa, and it supersedes the
later.
I
t consists of four chapters. The frst deals with the analysis of the prmazas
in general, te second with prtaka as prmaza, the tird with inference
(svarthtnumana) , and the fourh wit syllogism (pararhtnumana) . The next
work of Dharmakli is Pramala Viniscaa. It is an abridgment of PramalJa
Vtrika
.
The third work is Naa Bindu, which is a frter abridgment. The
remainig four works are small tracts devoted to specifc topics. Hem Bindu is a
short classification of logical reasons. Sambandha Parfkt is an examination of
the problem of relations. Codana PrkarlJa is a treatise on the act of carryig
on disputation. Santtnantar Siddhi/ is a treatise on te reality of other mids. It
atempts to refte solipsism.
Chapter 1
Te Buddhst Teor of
Kowledge
Every school of philosophy in India has attempted a teory of knowledge (rma{la
sastra) on which its metaphysical structure is built. Though te ultimate ainl of
philosophizing is generally recognized as knowledge of reality (tattva-iana) , it
is believed that a theory of knowledge is a necessary prerequisite to a teory of
reality. This belief is grounded in the fact that to philosophize is to refect on te
nature of reality given i experience. But since every experience is a cognitive
reference to an obj ect, tere is always te possibility of going astray in tis
reference. Though every experience has a built-in trasphenomenality or self
transcendence, it is not always guaranteed that it would adequately and faithflly
reveal its object . This possibility of error in experience necessitates an inquiry
into its veracity. In fact, te entire epistemological pursuit begins with, and
centers around, this task. Thiers belonging to te Buddhist tradition also,
therefore, attempt to provide a firm epistemological basis for their theory of
reality.
AAYSIS OF KOWEDGE!
"All successf human action is necessarily preceded by knowledge."
2
Wt
this prefatory remark Dharmaklri defmes te scope and aim of epistemology
and logic in the Naa Bindu. Human action may be either purposive or
instinctive. Human purpose is, again, something either desirable or undesirable.
A purposive action based on kowledge consists in attaining the desirable and
avoiding te undesirable. Knowledge is efficacious in causing successfl action
i te sense that it results in te atainment of te desirable ai and avoidance of
te undesirable one. A cause may be productive (kark) or informative (iapaka) .
Knowledge is a cause of successfl action in te latter sense only. It enables us
to reach te real, which alone has practical efficienc.
12 Buddhist Epistemology
Different from knowledge is false cognition. Cognition tat makes us reach
an object different fom te one revealed in cogntion is false. Objects differ on
account of their form and spatiotemporal locations. Thus, cognition representing
one form of the object is not to be considered as true when the real object has a
different form. Likewise, a cognition is not true if it wrongly represents te place
and time of the object.
In the Prma{la Vartika Dharmaklrti defines knowledge as a cognition that
is not in discordance with its object (Pama"lam avisamvadijfanam) .
3
He frter
maintains that a cognition that is perfectly in accord with its object will also be
characterized by novelty (Ajfata artha prktio va) . 4 It is revelatory of an object
not yet known because te object is momentary in nature and ony tat knowledge
will accord wit the object which arises at tat very moment when te object is
also in existence. It is the first moment of cognition, te moment of tle first
awareness. Continuous cognition is not valid or true, according to te Buddhists,
because it is not a new cognition but a recognition. Of these two characteristics,
te frst, nondiscordance, is basic and can be regarded as te differential character
of knowledge.
Knowledge thus stands for cognition tat is a faithfl representation of the
real. What it meas is that in knowledge te object must be known a it is and not
other than what it is. Since a nondeviating reference is an essential condition of
knowledge, the truth of kowledge consists in its accord wit the cognized object.
Dharmottara puts it as follows: "In common life when we say tat truth is being
spoken what we mean is that it makes us reach an object. Sililarly, that cognition
is true which makes us reach an object it points to. In fact, knowledge does not
create an object and does not ofer it to us, but just makes. us reach at it. By
making us reach an object nothing else is meant tan attending to it. 5 Here
Dharmottara points out three distinct, successive stages involved i te process
of apprehension of a object, each succeeding one resulting fom te preceding.
They are cognizing (adhigati) , attending (rvartana) , and reaching (rapa"la).
He makes it clear that the frst stage alone is knowledge.
Knowledge is of two types. It is intuitive when it springs from inside. It is
discursive when it is acquired by directing our attention toward an object wit the
help of te senses and the cognizing consciousness. Only discursive knowledge
is analyzed in epistemology.
T PROBLEM OF PRMA
A stated earlier, kowledge is a nondiscordant cognition. This means tat
not all cognitions are kowledge. Only those cognitions whose nondiscordance
is evidenced can claim te stats of kowledge. The problem of prmaza tat has
given rise to much stimulating debate in te epistemological treatises of Indian
origin is basically a problem of evidencing the truth of a cognition. The question
of evidencing a cognition arises because all cognitions are unequal i teir epistemic
status. Some appear to be true and reveal teir corresponding objects as tey are,
whereas others seem to be erroneous and misrepresent their objects. Had all
The Buddhist Theor of Knowledge 1 3
cognitions been true, there would have beep no need of evidencing them, and
the entire epistemological inquiry would not have arisen. The very possibilit of
error. in a c
o
gnition necessitates its subjection to a critical examination with a
view to establishing. its truth or falsit. If the trth or falsit of a cognition needs
to be established, the question arises what sort of criterion is to be resorted to.
The problem of prmtla has been discussed and raised precisely against this
background.
PRMAPHLA
Indian thinkers generally adopt a causal approach to kowledge. Knowledge
is taken to be an outcome of a particular causal complex in which the most
eficient instrumental cause (karla) is technically kown as prmtza. In the
Buddhist tradition, the word prmtlJa refers to both the process of kowing and
the kowledge acquired on that basis. Buddhists do not entertain the distinction
between the process of knowing (pramtlJ a) and its outcome
(prmtlaphaZa =prmt) . Whether or not prmtlJa and prmtlJaphaZa are to be
sharly distinguished has been a hotly debated issue between Nyaya and Buddhist
thinkers. Nyaya thikers insist that prmtza as a process leading to prmt should
be distinguished fom the latter, which is its phaZa (result). For them prmt is
the prmtlaphaZa, and prmtla is the karza of prmt. Buddhist thinkers,
however, maintain that no distinction can be possible between the noetic process
and its outcome. The act of cognizing completely coincides with the cognition of
an object. In fact, Naiyayikas have to accept the distinction because for them
prmtlJa is the evidencing condition for the truth of knowledge, whereas
prmt is the evidenced kowledge. For Buddhists, knowledge is self-evidencing,
and therefore there is no need to distinguish between prmtlJa and its phala.
Thus, the difference between Naiyayikas and Buddhists is due to the difference
in their understanding of the nature and role of prmtlJa. Naiyayikas understand
prmtlJa as that which is the most efficient causal condition giving rise to, and
evidencing, the kowledge (rmtytl karlam iti prmtlJam) , whereas for
Buddhists, it means that by which an object is known (rm/yate anho aneneti) .
Dinaga, however, does not refse to draw this distinction if it is needed
from a fnctional point of view. He writes, "We call the cognition itself prmtla
because it is generally conceived to include the act of cognising although pri
m
arily
it is a result." Commenting on it, he writes, "Here we do not admit, as the
Realists do, that the resulting cognition differs fom the means of cognition. The
resulting cognition arises bearing in itself the form of the cognised object and is
understood to include the act of cognizing. For this reason it is metaphorically
called prmtilJa although it is ultimately devoid of activity.
,,
6
Nature of PriapbaJa
In the DiTnaga-Dharmaklrti tradition two different views are available about
the nature of prmtwphaZa. According to one, prmtlJaphaZa consists in the
14 Buddhist Epistemology
cognition of an object (viaadhigama) . According to the oter, it consists in
self-cognition (sva-sarvitti) . Both these views are complementary and not
conficting and are available in the works of Diaga. 7 They are explicitly restated
by Dharmaklrti and Dharmottara. 8
S
antarakita gathers them together and brings
out teir distinction. He maintains tat according to the Sautrantia tradition,
which believes in bthtnhawda (realism) , strpa (similarity in forl) obtaining
between a cognition and its object is to be treated as prmtlJa, while cognition of
an object (viatdhigati) is to be treated as pramtlJaphala. According to
Viftnavtda, str pa is, of course, the prmtlJa, but sva-smivedana or sva
smivitti (self-cognition) is te prmtlJaphala. In the utinmte analysis these
views are not different because viatdhigati and sva-samvedana are not two
different phenomena as tey are two facets of the sanle knowledge.
8iripy as Pramia
It has been one of the perennial problems of philosophy to put forh a criteron
of truth that is cogent and convincing. Since kowledge consists in a cognition's
being fee fom discordance wit its object, the criterion of trth has to be
formulated in terms of kowledge's being in accordance with its object. Buddhist
thinkers therefore put forth str pya as a pramtlJa, in the absence of which no
cogniton can be taken to be knowledge. The doctrine of sarapya has been
forulated keeping in view the self-evidential character of knowledge and also
the view that since every knowledge is an awareness of an object, the conditions
of its truth are to be determined in terms of its reference to te object.
The Concept of 8iripy in Early Buddhist Philosophy
The doctrine of str pa advocated by Diiaga and DharmakIrti has its roots
in the Sarvastivada tradition,
9
tough it can frter be traced back to tlle early
Buddhist literature in PaL 10 I te Sarvastivada tradition idependent and objective
existence of the external object was admitted. Sinlilarity (strpa) of te form
represented in a cognition to tat of te object is held to be the prmtlJa of the
resulting cognition of t at object (viaadhigati) . Accordingly, a cognition tat
is produced along wit its object is true with reference to tat object. The
object, the cognitive sense, and te cognizing consciousness are all sinmltaneous
and momentary. All three exist at te sanle tinle (sahabhu hent) and constitute
tlle collocation of causes (ktrlJa-stmagn) tat gives rise to the cognition. This
Abhidharma accout of the genesis of perceptual cognition has been represented
by Stcherbatsky in tlle form demonstrated in Figure 1 . 1 .
The Buddhist Theory of Knowledge
Figure 1.1
Saabhu hetu
visaya indrlya viiana = manas
. - -
-
-
..
-
..
::
~ ~ ~ w meeting point of the three
15
(ttim sa
n
iatJ = pret cogton).
There the perceptal cognition is regarded to be a resultt of three tpes of
causal factors, namely,
1 . samanantara prtaa (i.e., perceiving consciousness),
2. adhipati prtaa (i.e., cognitive sense), and
3 . t[ambanaprtaya (i.e., object).u
Here a question naturally arises as to why a particular cognition should refer
to the object only for its determiation and not to the cognitive sense or the
cognizig consciousness, which are equally simultaneous with it. This is explained
by putting forth the doctrine of sarpa, according to which, though all three
causal factors are simultaneously present, the object alone provides a form to the
cognition and tus determines it.
The preceding view is put forth by the Vaibhaika school but it is not acceptable
to Sautrantikas, who reject the notion of "sahabhu hetu, " that is, the simultaneity
of obj ect, cognitive sense, and cognizin consciousness. The basic objection
raised by them against tis position is tat in the background of momentariness it
poses a serious problem. When the cognition arises, the object is no longer
present because, being momentary, it must have by then been destroyed. In such
a situation how can it be called the perception of tat object alone, and, frter,
how can that object be the determinant of the truth of the cognition? The
Sautrantiks therefore solve this problem by putting forth the doctrine of sarpya,
according to which the object no doubt is momentary but leaves its impression
upon consciousness, and through tis impression, which has sarpa with the
object, the object is cognized.
12
Though bot the Vaibhaiks ad Sautantiks accept te doctrine of sarpa,
they use it difrently. The forer use it to explain why the cognition is determined
by the object and not by te cognitive consciousness, though both are the causal
factors giving rise to the cognition; while the latter use it to explain how the
object that is already destroyed is cognized in perception.
The Doctrine of Saripya in the Diiaga-Dharmakiti Tadition
In the Dilaga-Dharmaklrti tradition the doctrine of sanlpya has been
primarily put fort as a prmaza for perceptual cognition, and in this sense it
1 6 Buddhist Epistemology
marks a departure fom te Sautrantika use of tis concept. In fact, Stcherbatsky 13
has pointed out three different uses of the term santpa, namely:
1. to refer to the relation between a sensation and a conception,
2. to refer to the relation between syalaafd and its prtbhtsa, and
3 . to refer to the relation between a svalakafa and its corresponding stmtna laafa.
In the Diiaga-Dharmaklrti tradition it has been used i te second sense.
TH DOCT OF SY-PRKSA (SELF-RVELATION
The doctrine of sarpa has been put fort in the context of the more basic
doctrine of dvairia, which implies that every cognition is necessarily in the
fonn of a twofold appearance, namely, that of the object and tat of itself. This
doctrine of dvairpya has been an inevitable corollary of te doctrine of sva
prkt (self-revelation of cognition) , which results in the obliteration of the
distinction between prma{lQ and prmalJaphala, insofar as the same cognition is
at. once both the evidence and te evidenced.
The doctrine of sva-prkasa has been the basic tenet of the Buddhist
epistemology put forth in the context of te teory of momentariness. If a cogntion
is just a momentary state of existence ceasing to exist the next moment, then
either we can have a cognition of that cognition in the very moment of its origin,
or we shall have to deny the very possibilit of the cognition of that cognition
because the next moment that cognition will no longer be there to be cognized by
the subsequent cognition. Since the cognition of a cognition is a fact given in our
experience, it can be understood only in the context of the doctrine of sva
prkasa. Consistent with the theory of momentariness, Buddhists talk of the
cognition of cognition only within the framework of simultaneity and never that
of succession. Thus, the only position available would be to advocate the theory
of sva-prkasa, which would mean that at the very moment of the cognition's
cognizing an object, it also cognizes itself.
Diiaga and Dharmaklrti use the expression "svasmivitti " to express this
fact. 14 Thus, when a man has the cognition of something blue, he has at the same
time the awareness of the cognition of something blue. This awareness is caused
by nothing other than the cognition itself. Thus, the cognition, while cognizing
an object, cognizes itself just as a lamp illuminates itself while illuminating an
object.
The Doctrine of Dvaiipya
The doctrines of sva-prkasa and svatafl-prama(lya (self-evidentiality of
cognition) go together. Not only is every cognition a cognition cognizing itself
but it also evidences itself. Now te question is, How is it tat a cognition cognzes
itself and thus establishes itself as true? To answer this, as stated earlier, the
doctrine of dvaintpya has been put forth. 15 Every cognition is produced with a
t wofold appearance, namely, that of itself (svabhasa) and that of the object
The Buddhist Theor of Knowledge 17
(viaabhisa) . In being viaabhisa i t establishes its trut because of its being in
te form of the object.
1
6 When a cognition possesses the form of an obj ect, it is
a sufcient condition for its being true. Thus, Dilnaga writes, "The pramala is
simply the cognition having the form of the object. "
1
7
Hatori
1
8 has made an attempt to represeit the doctrine of dvainpa using
the following symbols :
Viayajiana = C
1
Svabhisa in C
1
= S
I
Arthibhisa in C1 = 0
1
Viaajiana-jiana = C2
Svabhisa in C2 = S2
Arthibhisa in C2 = 02
He writes tat, according to Di naga, C1 = (S
I
- 0\ ), C2 = (S2 - 02) ' ( "-"
indicates relation) . Since C2 takes C1 for its object, 02 = (SI - 0
1
) ' Therefore,
C2 = {S2 - (SI - 01 )} ' Thus, C2 is distinguishable fom Ct
Now, i case te cognition had only arthikara ( = abhisa) , then C
1
= 01 _
and C2 = 02' Since C2 takes C
1
for its object, 02 = 0\ . Therefore, C2 = Ct .
Thus, viaajiana-jiana would be indistinguishable from viaajiana. If, on the
other hand, the cognition had only svakara (= abhasa) , then C1 = Sl
'
and C2 =
S2 a However, sice the cognition, which does not possess the form of an object
withi itself, remains te same at all times, Si = S
I
' Therefore, C2 = Cr '
In fact, Hattori goes beyond C2 to postulate C3 ad C4 and so on ad infnit.
He writes, "In the same manner, the tird and the succeedig cognition C3 ' C4
. . . . . . Cn are distinguishable fom the preceding ones only when they are
ackowledged to possess "dvi-npa. The formulas are as follows :
(S3 . . Sn and 03 . . . . On' respectively, stand for svabhisa and arhibhi sa
in C3 . . . Cn, C2' C3 . . . Cn a conttut b addig aoter am t te prg CI '
C2
. .
Cn
_
j

"
This symbolic representation and its explication given by Hattori do not
seem to be very appealig, mainy because tey are based on te presumption
that i. te self-cognition of the cognition, which ivolves two cognitions, namely,
Cj and C2' C
1
is the precedig, and C2 is te succeeding one. Hattori proposes
to go beyond C2 to C3 ' C4 ad ifnitm. In tis venture he seems to be working
under te " succession-model . " But as we have stated earlier, Dilaga' s teory of
sva-prkasa canot adequately be understood like tat. Dilaga is quite explicit
in advocatig te " simultaneity model " which would preclude any overstepping
from C
1
to C2. If C
j
is granted the dual form of svakar and arthikr, which
1 8
Fgure 1.2
Drr pyModel
Object (viaa)
I
cognition (itna)
I
svtktra (rmtla-phala)
Buddhist Epistemolog
arhtktr ( ramla)
I
(arha-strpa) strpa
arhthigati svasamvitti
sva-prakta teory pronouncedly itends to do, tere is no need to bring in
C
2
,
and if
C
2
is needless, we also have, ipso facto, not to go beond
C
2
to
C
n
. Thus,
a distinction between
C
1
( viaajitna) and
C
2
( viaajitna-jitna) would
violate te very spirit of Diiaga's teory of sva-prkta. In fact, witin
C
1
itself we have svtbhtsa and arhtbhtsa, which are inextricably so interspersed
tat tey can be distinguished only for te sake of conceptual clarity but can
never be separated. However, tis distinction does not in aiy wy imply tat
svtbtsa (viaajitna-jitna) and arhtbhtsa (viaajitna) are two diferent
occurrences. In te very act of kowing an object tere is an awareness of the
kowing becase had tere been no awreness of te kowing at te ita moment,
it could not be possible to have it subsequently, as neither te object nor te
cognition of the object would last for te next moment.
Hattori seems to understand te Buddhist teory of dvairpa as implying
that at the frst moment there is a rise of the frst rpa, namely, tat of
viatktra, and then at te succeeding moment tere will be te rise of the
second rpa, namely, tat of svtktr, but such an iterretation doesn't do
justice to Diiaga' s position, which explicitly maintains tat a cognition has
,
svtktra along with arhtktr ( viatktra) .
The Buddhist teory of dvairpa understood in te context of "simultaneit
model" is represented i Figure 1 . 2.
T CONCEY OF VYA
The Buddhist teory of dvairpa as represented in Figure 1 . 2 states clearly
that arhtktritt, also kown as arha-strpa or just strpa, is te pramtla of
which arhtdhigati (apprehension of te object) and svasamvitti (self-cognition)
are te prmtla-phala. There has been a controvers among Buddhist logicians
preceding Diga as to wheter arhtdhigat is te praml-phala or svaavtti.
Of course, tis controversy presupposes the distinction beteen pramtl and
The Buddhist Theory of Knowledge 1 9
prama{laphala for pragmatic puroses ony. Granting this distinction, Diiaga
writes that, strictly speakig, arthadhigati alone is the prrafaphala, but
svasamviti can also be regarded as prmafaphala.
19
Every cognition is a cognition of something and ths inevitably refers to an
object. The object gives rise to, and thus determines , the cognition. The object
is regarded as the alambana prtaya, that is, the very ground upon which the
cognition is based. In fact, the object not only serves to give rise to the cognition
but also differentiates it fom other cognitions. Without assigning this task of
differentiation to the object, one cognition canot be differentiated fom another
cognition. Though from the transcendental point of view every phenomenon is
viiaptimatr (pure consciousness) , at the empirical level a triple division of
viiana (consciousness) is drawn ito grahyakar (the form that is cognized) ,
grahaka (te cognition) , and svasarhvitti (self-cognition) . The grahyakar serVes
as a diferentiatig factor between one cognition and anoter.
2
0
Since every cognition is determined by the object, tis determiation should
be understood as the cognition's having the form of te object. Thus, if the
object is a pen, the corresponding cognition should have te form of pen. Then
only will it be said to be determined by the object "pen" and be a te presentation
(abhasa) of that object. If the cognition is at variance with the object, it will not
then be determined by the object, and this will amount to its falsity. In order that
a cognition should be tue, it has to refect or represent the object in its real form.
This will be possible only when the cognition is arising in the form of the object.
The truth of a cognition, therefore, consists i this sameness of form wit the
object . It is not that knowledge is formless , as is maintained by the
nirakrjianavadins. If the cognition were formless while the object had a form,
then the cognition itself as distinguished fom te object would remain the same
whether it cognizes something blue or yellow or anythig else. Therefore, the
cognition as the apprehension of the object must be admitted to be sakar. To say
that cognition is sakar is the same thing as to say that it has the form of the
object. The sakar cognition is thus understood to possess the fnction of assumig
the form of the object. For this reason Di aga considers it as prma{la, although
primarily it is a phala i its aspect as apprehension of an object.
2
1
.TA A A.TA PRTIHSA
Accordig to Buddist tiers sice every true cognition (and, for tat
matter, every cognition) is necessarily caused by a object, and since tere are
only two types of objects, naely, te graha (given) ad te adhavasea
(construed) , tere are two types of cognition, te perceptual and te conceptual .
Corresponding to tese are two types of prtbhtsa (menta refex) . The perceptua
cognition consists i a prtibhtsa tat is niata, tat is, defmite and not vague,
because it is caused by a object tat is independent of te perceivig consciousness
ad terefore objectively real . In case of conceptual cognition tere is no such
objectively existing real object, ad hence te prtibhasa here is aniata, tat is,
idefiite ad vgue. Te apprehended object givig rise to a awreness generates
20 Buddhist Epistemology
a regulated mental refex (i . e. , a refex liited to tat very object) . It is just like
a patch of color that, giving rise to visual awareness, generates a defmite mental
refex liited to tat patch. The conceptual awareness, on the other hand, is not
generated by the object. Thus, in te absence of the causal factor to regulate the
mental refex, tere is no fxed (defmite) mental refex. However, i should be
made clear that fom the empirical point of view tis aniyatatva (indefmiteness
of pratibhtsa) in no way afects the epistemic status of inferential cognition
because the vlidit (rama1)ya) of a cognition depends not on the vagueness or
defmiteness of prtibhtsa but on its sarpa (conrt wit te objet) , ad i te
ierntal cogton it i ver much prsent.
Diiiga's Analysis of Prtibhasa
Since the object gives rise to its prtibhaa, te pratibhasa has to be in
coordination with its respective object. According to DiInaga, every pratibhaa
has to be in coordination with its object because it is produced by that object
only. A object canot produce any such prtibhdsa that cannot accord wit it.
Logically, therefore, he rules out any possibility of deviation from its object in a
prtibhasa.
Dharmakrti's Analysis of Ptibhasa
Dharmakrti, however, maintains tat sometimes an object may fail to give
rise to its genuine pratibhtsa. This may be due to several factors like the locus of
the object, the circumstances of te presentation of te object, the disturbed
state of cognizing consciousness, and so on. If a prtibhtsa fails to accord with
its corresponding object, it will iack coordination wit it, and to that extent it
cannot be said to be genuine. Thus, when one has the prtibhasa of a moving tree
while traveling in a boat, the pratibhtsa cannot be said to be genuie because it
is not in accord wit its object. Of course, such an analysis of te epistemic
situation would not be acceptable to DiIaga, who would insist tat every
prtibhasa has to be in accord with its object because it is produced by that
object only.
Irrespective of whether a prtibhtsa may or may not deviate fom its object,
te fact remains that every prtbhasa i a true cognition must be in accord wit
its object . That is why Buddhist thinkers insist on avisarlvadakatva
(nondiscordance) as a necessary condition of a true cognition.
Is There Arhasaripya i Iferential Prtibhasa 7
Wile commenting on te Naya Bindu of Dharmaklrti, Dharmottara seems
to maintain tat tere can be only one tpe of real object, namely, te svalaka1)a
(unique particular) . 22 There is no real object corresponding to te conceptual
cognition insofar as te samanalaka1)a (generalized image or concept) is not
objectively real . This will ten mean that in a conceptual cognition tere will not
be any possibilit of arthasarpya (coordination) wit te really existing object.
But Dharmottara
'
s contention can be sustained only if we overlook te fact tat
The Buddhist Theor of Kowledge 21
for Buddhist tinkers samanala"a1a i s also real, altough, of course, i is real
only from the empirical poit of view (sarv{ti sat) .
Justifcation for Accepting S;r py
It seems that te
o
nly reason for the Buddhist tiners to advocate the teory
of sakr ji na is to provide for te determination of te cognition b its respective
object. Every cognition refers to an object insofar as it is produced b te object.
There is no cognition tat is not produced by te object and tat terefore is not
having some form given by the object. Thus, tere is no formless cognition,
because in te very process of its being produced b te object, it gets, so to say,
the form of the object. It may be tat te for of a cognition does not accord wit
the specifc object tat is given i the epistemic situation and may accord wit
some oter object tat is not given but only hypostatized. Such is the case in
illusion. Here, there is no arthasarpa, and hence it is not a true cognition. The
cognition here has a form but not of te given object. Thus, te doctrine of
sarpa serves te purpose of distinguishing beteen a true cognition and an
erroneous one. Buddhist tinkers uderstand tt i ters of avsawtvam,
tat is, nondiscordance wit the object of cognition. The necessary as also the
suffcient condition for te availabilit of nondiscordance is arhasarpa, which
means te cognition should be determined by tat very object and not by any
oter object. If it is deterined by some object oter tan its own, it will be a
case of vairopa (discordance) rater than tat of sarpa, because in that case
some oter object will be te determinant rater than te legitimate one.
Having analyzed the doctrine of sarpa, we may proceed to see te grounds
for its acceptance. Though this doctrine is deduced as a corollary of te doctrine
of dvairpa, t grounds one positive and te oter negative have been put
fort to account for its acceptnce. The frst ground is te fact of recollection,
which provides te basis for saying tat te cognition has two forms and tat it is
in accord wit its object. A DiInaga puts it, "If te cognition has only one
form, either tat of the object or of itself, ten te object which was cognized i
te preceding cognition could not appear in te succeeding cognition, because
tat object of te preceding cogntion would not exist when a succeedig cognition
arises. But we do have facts of recollection when some time afer there occurs to
our mind the recollection of our cognition as well as tat of te object. We
express te fact of te recollection of a past cognition i te form ' 1 remember
that 1 cognized te object.' The recollection of a past cognition is explainable
only by admitting that the cognition is cognized by itself. "
2
3
The second ground for admitting sarpa as a prama1a is tat if a cognition
were not self-cognizing and self-evidencing, tis would have resulted in infmite
regress, because it would have required anoter cognition for its cognition and
validation, and te second one would have required a third one, and so forth.
The doctrine of sarpa is flly consistent wit te idealistic position of
Vijianavada Buddhism, for which in te ultimate analysis all objects tat appear
to be objective and external are nothing but projections of the consciousness.
22 Buddhist Epistemology
Because of this the Buddhist can talk of the cognition appearing in the fon of
obj ect. A has been pointed out earlier, the idea of cognition assuming the fonn
of the obj ect should not be mistaken as sameness of fon between the obj ect and
the cognition. It stands only for the detenination of the cognition by the object.
However, the ten sarupya is slightly misleading insofar as it gives the impression
as if cognition has te same or similar for possessed b the obj ect. It is meaningfl
to talk of the fon of an object, but it doesn't make much sense to talk of the
fon of a cognition unless it is understood i a fgurative sense only. All that it
should be understood to mean is that every cognition necessarily refers to an
object, and ever tre cognition has to refer to the object as it is. To refer to the
obj ect as it is means to be of the fon of the object. Understood in this way the
contention of the Buddhist seems to be plausible, but to express this with the help
of the ten "saripya " is beset with the difficult referred to before. In fact, the
Buddhist logicians should have used another word that would have better expressed
the idea of the cognition's being detennined by its corresponding obj ect.
TORY OF MYADHAMASIDH
The idea that te trut of a cognition is to be determied in terms of its
reference to its obj ect leads Buddhist tinkers to te teory of Meadhina mana
siddhi . According to DiInaga, knowledge is no doubt revelatory of its obj ect,
and therefore object depends on knowledge for its revelation; the truth of
knowledge, however, depends on its exct reference to its object. So far as epistemic
dependence is concerned, te object depends upon knowledge, whereas wit
regard to ontic dependence, knowledge depends on te object. In te context of
trut antic dependence is taken into consideration by Buddhists. However,
Naiyayias sharply differ from Buddists and insist that though knowledge is a
faithfl awareness of its obj ect, te trut of knowledge is not determined and
guaranteed by tle object but by te most efficient causal condition, which is
technically known as kar1a. In fact, this is a natural corollary of the Nyaya
distinction between prmala and prma1aphala. TlS, for Naiyayikas all tat
is real is kowable, and te trut of knowledge is generated (utam) and vouchsafed
(iapti) by prnl(lG only. This is known as the teory of ManMhinameasiddhi(1
In other words, for Naiyayikas epistemology determines ontology, but for
Buddhists tle reverse is the case.
OBJCT OF KOWEDGE
According to Buddhists, there are only two kinds of objects of knowledge,
namely, the real, objective, unique particular (svaZaka1a) and the generalized
concept or image (samanaZak$ala) . This is because the thig to be cognized has
the preceding t wo aspects. The objective real i the fon of unique particular has
no extension in space and no duration in time. It is devoid of all form, attributes,
deteniations, and relations. It is just a poit instant or a moment in the incessant
moments of series of reals. It is unique particular in the sense that it is neither
identical wit, nor similar to, any other object. It is a distinct existence dissimilar
The Buddhist Theor of Knowledge 23
to, and noncomparable with, any other real. It is instantaneous because i never
endures for the next moment. It is discrete in the sense that it is an isolated
existen<e not
'
at all related to any other existence. So no relation exists between
any two unique particulars, all relations being subjective mentl constrctions.
Such an object alone is objectively real. This is a very distinct and unique view
held by Buddhists with regard to the nature of reality. Reality is essentially
momentary and dynamic. It is a process i which each moment depends for its
existence on a previous moment. Each moment is in itself unique but causally
determied by the previous moment. Being momentary, we canot assign any
name to this reality because the act of giving a name implies that ( 1 ) te real has
to exist for more tan one moment in order that a name could be recalled and
associated with it and that (2) since the fnction of naming requires the possibility
of identifig a thig whenever its name is uttered, a momentary real canot be
named, and whatever is named is not the reality proper. Therefore, tere is one
aspect ofreaIit-in-itself, and tere is another aspect of reality tat is conceptualized
and talked about in general terms.
Thus, distinct from the unique particular is another type of object of
kowledge, kown as sanlnalaka1a, which is a construction of our mind and
which is i the form of a generalized inage. The generalized image is a form
imposed by our mid on the objective reality. A generalized image, also kown
as concept (vikalpa) , can be, broadly speaking, of five types, namely, pertaining
to substantivality (drva) , adjectivality (gU1a) , relations and spatiotemporal
locations (kanna) , class-character (ati) , and linguistic determination (nama) . A
generalized image is a mental construction (kalpana) having no objective existence.
Comprehension of objects as having extension in space and duration in tinle is
nothing but generaliation, which is only subjective or intersubjective and has no
counterpart i the objective world. Siilarly, all attributes, relations, and so on
are nothing but generalizations.
KIS OF KNOWEDGE
On the basis of the preceding ontological analysis Diiaga enunciates the
theory tat since there are only two kinds of objects of kowledge, there are only
two kinds of kowledge. Corresponding to the unique particular (svalakWlQ)
we have perceptual kowledge (rtaka) , and corresponding to the generalized
iage we have iferential kowledge (anu1ana) . A kowledge is either perceptual
or inferential, and there is no kowledge beyond te purview of these two. Thus,
the entire Buddhist epistemology is based on te foundations of the theory of
twofold kowledge. Perception (rtaka) is pure sensation, a direct sense
apprehension of te unique particular. Inference (anu1ana) is a mental construction
i the foml of generalized iages. Of course, perceptal kowledge is imediately
followed by te inferential one; te fomer is not at all judgmental or determiate.
Buddhist thiners emphatically maintain tat te unique particular is kowable
i perception only and tat te generalized iage is kown only through inference.
24 Buddhist Epistemology
By implication te unique particular can never be known in iference, and likewise
the generalized unage can never be known in perception. What is known in
perception camlOt be known in inference, and viceversa. Such a position of
radical dichotomy of mutually exclusive modes of knowing is known as
prmcavavastht, which means that each of te two modes of knowing has its
own separate and distinct sphere of operation.
There is no intermingling Ul the respective objects of the two. A the unique
particular alone is objectively real, while te generalized liage is a mental
construct, and the one is radically diferent from the other, there cannot be any
cognition that comprehends both at te same tinle.
Chapter 2
Te Buddhst Teory of Perception
In the Buddhist theory of knowledge perception (prtaka) is regarded as
the foundational prmala insofar as inference (anumana) depends on it .
Inference presupposes perception, and, therefore, only in this respect does
perception acquire primacy over inference . Though all Buddhist thinkers
fom Dirnaga onward agree on this point, their understanding of the natre of
perception is not the same, and hence they defne it in diferent ways .
DEFIION OF PRTAK
To give a logically concise and comprehensive defmition seems to be a very
difcult task. Some tinkers may maintain a extreme position tat such a defmtion
can never be given. If the thing is kown, tey may argue, its defmition is
useless, and if it is not kown, it is still more useless because it is impossible.
This, of course, has not been te view of te Buddhist thinkers we are concered
with. Among these thinkers Diimaga not only puts forth definitions of
philosophical terms in as sharp and clear-cut way as possible but also gives a
specific metod of defming with te help of double negatives (atadvav
.
rt i) .
Nyaya-Mlasa thinkers attempt to define tings in terms of teir "essence, "
tat is, their characterizations and relations. Dilaga, on the contrary, contends
that such defiitions are useless because te "essence" doesn' t exist. Thus, things
qua-things (svalakal)as) are indefinable. However, our conception of a ting
(vikalpa) is definable. According to hinl , the characteristic feature of all our
conceptual knowledge (and of language) is that it is dialectical . Every conception
is a negative corelate of its couterpart and while defming a concept tis alone
can be stated. So, a definition is only a negative characterization (vavrtti) . For
example, what te color "blue" is we canot tell, but we may divide all colors
26 Buddhi st Epistemology
into " blue" and " non-blue. " The "non-blue" may mean any color oter than
blue. The defnition of "blue" will be "tat which is not non-blue, " and conversely,
te defmition of " non-blue" will be "tat which is not blue. " For Buddhists all
lowledge can be dichotomously classifed into direct and indirect. The direct is
not indirect, and indirect is not direct. The direct cognition is techically lown
as prtaka, and te indirect as kalpana or paroka. Prtaka can terefore be
defied i terms of its distiction from kalpana or paroka. From tis it follows
that prtaka can be understood as the " negative of kalpana. "
TEE WAYS OF DEFIG PRTAK$
In the later Buddhist epistemology tree distinct positions are discerible in
an attempt to defne prtaka. Though there seems to be unanimit among
Vasubandhu, DiIlaga and Dharmakrti tat te defnition of prtaka should be
in terms of " negative of kalpana, " they sharply differ with regard to the
phraseology tat is to be used to express this point. Vasubandhu does it by using
positive phraseology. Thus, he falls on one side, while DiInaga and Dharmaklrti
on the other. The latter, though they use negative phraseology, differ between
themselves not only about the fll rendering of the phraseology but also about
the epistemic status of prtaka. We first state te three positions in detail and
then undertake their comparative analysis, particularly of the positions of DiInaga
and Dhannaklrti.
VASUANH' S DEFION OF PRTl
On the basis of the inforlation available to us Vasubandu can be regarded
as the first systematic epistemological thier pertaining to te tradition followed
by DiIaga a Dn. I the Yd-Vidi pmtaa i defed a " tto 'rhi d
vjfnm prm, "
1
that is, " perception is a cognition produced from that
object. " In this definition Vasubandhu maintains tat prtaka is a cognition
that is deterlined by its object. Here the word "tato" points out te exclusive
role of the object (alambanaprtaa) in contradistinction to the role of the
cognitive sense (adhipatiprtaa) o
r
the cognizing consciousness (samanantara
prtaya) .
2
For example, if the object is fire, which is an existing real
(svalaka{w) , and if it is cognized, ten cognition of fire is a case of perceptual
cognition. Such a cognition has a real object, which is fre, as distinct fom the
conceptual object, which is te idea of fre. Bare experience of the object without
any conceptal element is regarded as perceptal cognition. This differential
character of prtaka, which consists in being " negative of kalpana, " has been
expressed by Vasubandhu with te help of the phrase " tato 'rthad. " The idea to
be expressed by the phrase "tato 'rthad" is by iplication te same as " without
any conceptualization. " Therefore, Vasubandhu can be interpreted as defming
perception as bare experience that is solely and wholly caused by te object
without any intermingling of conceptual elements. Though Vasubandhu does not
The Buddhist Theory of Perception 27
specifcally say that a perceptual cognition has no element of conceptalization
and verbalization, in fact, he means it when he says that it is exclusively coming
from the object.
DIN AGAS DEFIITION OF PRTYK
Instead of defning prtak$a in terms of "tato 'rhad vifanam, " Diaga
'
defines it in terms of "kalpanapoham, " perhaps because the idea of prtak$a
as " negative of kalpana" can be better expre
s
sed by te latter phrase rather than
by the former. Afer having stated that there are ony t wo kinds of knowledge, 3
which he conventionall y call s prtyak$a and anumana, DiIinaga turns to
prtaka and says that tis kowledge is kalpanapoham
4
(nonconceptual) , which
is anoter way of stating that it is " negative of kalpana. " Earlier, he argued that
there are only two types of cognition and tat tere is no need to postulate tree
or more types; otherwise, it may lead to infinite regress.
s
Thus, the object can be
cognized either in pratak$a directly or in anumana indirectly; there is no oter
way of cognizing it.
pmaga seems to have tw objectives i m when he proceeds to defie
prtak$a, namely,
1. to distinguish it from anumana and so on (arravavrttartham) and
2. t o di stinguish t he Buddhist conception from t he vi ews of other schools
(viprtipattinirakarmJartham) .
Rejection of Nyaya-lama Defnition
The usual definition of prataka given by Nyaya-Mlmasa and other
non-Buddhist traditions in terms of sense-bjet contct (indriyarha-sannikar$a)
has been rejected by Diaga maiy because it takes no notice of te basic
feature of prtak$a, which is always a new cognition, tat is, cognition of
somethig new and not recognition. Such a cognition can be only i the form of
sense perception, which is the first moment of every cognition. In the following
moments when te attention is aroused and te process of mental construction
begins, it is no more tat pure sense perception tat it was at te first moment.
Moreover, tat definition contains a concealed confsion between te proper
function of sense perception and the fnction of te mind. Sense perception has
its own fction, its own object, and its own cause. Its fnction is to make the
object present to the senses (sak$ttkaritva vapar) , not, of course, forcibly (na
harMt) . Its object i s svalak$alJa (unique particular) sice tis alone, being real
and efcient, can produce a sense impression, and the svalak$aza (wique
particular) again is its cause. Thus, prtaka consists in an awreness (rtibhtsa)
of te presence of an object, its mere presence and noting more. To construct an
inlage of a object (rtibMsaprtl) the presence of which has tus been reported
is another function, a subsequent operation that follows in the track of the
first.
28 Buddhist Epistemology
Rejection of Vasubandhu's Defnition
Dilaga disagrees with Vasubandu' s way of expressing pratak$a in terms
of "tato 'rthid" mainly because it suffers from ambiguit. Dilaga, being an
abstrse logician, does not want to leave any lacuna i his definition tat he
detects in tat of Vasubandhu. In saying tat a perceptual cognition is tat which
is exclusively caused by the object, it is not clearly specifed whether or not tis
forbids the involvement of mental constrction. Dilaga states this point quite
explicitly by defning perception as kalpanapo{ham" (non-conceptual) .
Comparison of the DefInitions of Vasubandhu and Diimaga
In fact, tere does not seem to be any fndamental difference between
Vasubandhu' s defmition and that of Dilaga. The same fact has been positively
stated by Vasubahu and negatively by Dinaga. Both agree tat what we normally
take to be perceptual cognition is, in fact, a complex of perceptal ad conceptal,
a synthesis of the contribution of the object and of the knowing mind (cita) . If
we represent this complex as "K, " the contribution of te object as " S, " and the
contribution of the kowing mind as "c" and syntesis as " +, " then "K" can
also be represented as "s + C" (i . e. , K S + C) . According to both thinkers,
te total "K" is not perceptual ; it is a complex of perceptual and conceptual . Out
of this complex only " S" is perceptual , not " C. " This fact has been stated by
Vasubandhu in saying tat " S" is perceptal and by Dilaga in saying tat "K -
C" is perceptal. But, i fct, te two expressions, "S" and "K - C" are equivalent.
The linguistic expression corresponding to " S" is a little ambiguous, whereas the
linguistic expression corresponding to "K - C" does not contain any ambiguity.
Perhaps, Vasubandhu
'
s defnition could have been improved upon by the addition
of the word " eva, " which would have meant that a perceptal cognition is that
which is determined by the object solely and exclusively. But Diraga thinks it
better to put the defmition negatively so as to avoid any scope for ambiguity.
However, it must be made clear tat Dirnaga does not reject Vasubandhu
'
s
defiition but only brings out its implication i clearer terms. A perceptual
cognition is solely determined by its object; it must be wholly given by te object
and not in any way constructed by te mind. Thus, Dil naga specifies tat it is a
cognition that is not at all subjectively determined and conceived of in terms of
the vikalpas (concepts and words) of te tpe of drva (substace) , gua (qualit) ,
kamza (action) , jdti (class) , and nama (words) . What is perceived by us is unique
individual , which is bare existence, devoid of all characterizations. It is just what
is immediately given to us in experience. Concepts and words are common
(s{l1nanyalak$aa) to several objects, and they are not unique (svalak$aa) . From
this it follows that, according to Diraga, perceptal cognition is te imediately
given sensum in complete isolation from all conceptual determinations.
The Buddhist Theory of Perception
PERCEPTION AS DISTICT FROM ERRONEOUS COGNITION,
IFERENCE, A SO ON
29
One of t
h
e objectives of Dilnaga, as stated before, in defining perception
has been to distinguish it in clear terms from erroneous cognition, cognition
of apparent reality, inference, its result, recollection, and desire, which are not
perception because they are not immediate on account of being vitiated by
obscurity.
6
Explaining this, Di naga writes i te auto-comentary (sva-Vr l)
that erroneous cognition is not perception b
e
cause it arises due to conceptual
construction, for exmple, the cognition of water i a mirage. Cognition of apparent
reality is not a true perception, which is only empirically true. Inference and its
resultant cognition and so on are not perception because they superimpose
something extraneous upon tings because they arise tough te conceptualization
of what formerly has been perceived.
1
W T NAM PRTAK$ A NO PRTSYA?
Perceptual cognition arises when an object gives rise to its refex
(pratibhtsa) in the cognizing consciousness through the senses . Here
Dilaga poses a problem: "For what reason then is it called prtaka and not
prtiviaya, despite te fact tat it is dependent upon bot. " 8 To tis he replies
that it is named afer the senseorgans, because they are its specific cause. The
object is only a common cause because it is common to many cases insofar as it
is a cause of siilar perceptions i one' s own self as well as i oter person' s. In
fact, this problem is earlier posed by Vasubandhu,
9
who gives two reasons, of
which Dilaga mentions only one. The other reason is that as te cognitive sense
is strong or weak, perception becomes clear or dim, and because of tis, sense
organs should be regarded as the specific cause of perception.
From this account it is quite evident tat Di naga is the frst systematic
exponent of te theory that perception is strictly indeterminate (nirvikalaka)
and excludes all conceptual constructions and verbalizations (vikalpas) .
However, for him, every perception has to be nonerroneous (abhrtntam)
because all errors are consequences of mental construction. At the level of
pure sense data there is no question of error because the sense data are wholly
given by the object, and there can be no error here. Error arises only when the
mental faculty comes to work upon . the sense data. That is why he does not
include te term "abhrtntam" in his defmition of perception, as his successor,
Dharmaklrti, has done. In te Ntasutr of Gautama also, perception has been
characterized as avabhictri (nonerroneous) , but for Dilnaga vabhictr (error)
just cannot be tere at the level of sense data.
DEFIITION OF PRTAK$ BY DHiTI
After having stated that, there being two types of objects , there are two
kinds of knowledge, Dharmaklrti, following Dilnaga, defnes prtaka as
"kalpantporham" (nonconceptual) but, difering from Dinaga, adds another
30 Buddhist Epistemology
term, "abhrcmtam" (nonerroneous) to it. He defnes it intens ofa cogton that i
fe fom concet constcton ad err. lO
Object of Pratyaa
Perception, according to DharmakIrti, consists i the apprehension of an
object in its own specifc character (svalaka1a) , which has nothing in conmlon
with other objects similar or dissililar and is completely free from association
with names and verbal expressions, which are conceptual constructions. The
object of perception is thus te real , which is immediately revealed to te
consciousness and not such objects that are in te form of concepts and words
(vikalpas) , which are not a part of the given sense data but are conceived in te
mind.
The object of perception is different not only from te objects that are in
te form of concepts and words but also from a bare figment of imagination.
The unique particular (svalaka1a) alone is the real object, and, being the
real and efficient, it alone can produce a stimulus upon te senses. A figment
of inlagination, on the other hand, is unreal and, terefore, canot exercise
any such stimulus. Commenting on Dharmaklrti ' s opening remark in the
Nta Bindu tat " all successfl human action is preceded by right knowledge, "
DhamlOttara in te context of perception writes tat when an object is apprehended
by direct experience, it gets converted into an object of purposive action trough
sense perception. In fact, this characteristic feature of the real object distinguishes
it fom all objects that are absolutely unreal and that have no practical efcienc
(arthakriytktrita) .
1 1
We may now analyze the two terms, namely, "kalpantporham" and
" abhrcmtam, " that constitute Dharmaklrti ' s definition. By using the term
"kalpanaporham, " Dharmaklrti attempts to distiguish a perceptual cognition
from all such cognitions that are in the form of conceptual construction. He
defines kalpaTt as "abhiltpasmilsargayogaprtibhtsaprtfti, "
1 2
that is, a
determinate cognition tat is capable of verbalization. In this defnition there
are two aspect s of kalpana that must be expl icitly stated, namel y,
"prtibhtsaprtiti " and "abhiltpasamsargayogyata. " Every kalpant is a
prtibhtsaprtfti; that is, it is not just bare cognition (rtibhtsa) but a detemliate
cognition (prtibhtsaprti!) . To be aware of te bare existence of an object is
prtibhtsa but to recognize that object as "tis" object, a particular object, is
prtibhtsaprtii. The object is capable of giving rise to mental refex (rtibhtsa)
only. It canot produce determination in te form of recognition that "it is such
and such" , because corresponding to the determination (rtft!) tere is no object
present . The determination canot be regarded as a part of perceptual cognition.
Thus, Dharmakrti draws a clear-cut distinction between prtibhtsa and prtibhtsa
prtfti. The pratibhtsa alone is genuine perception and not prtibhtsa prtfti,
which follows prtibhtsa. The prtibhtsaprtlti is kqlpana, which is paroka
jfana o anumana.
1
3
The oter apect of kalpana is abhilapasansargaogatt.
The Buddhist Theory of Perception 3 1
Dharmakirti points out that a prtibhasa i s incapable o f verbalization
(anabhilapa) ,
14
whereas only a prtibhtsaprtfti can be expressed in language.
In oter words, perception, which is pure sensation, is only to be experienced. It
cannot be designated. I is interesting to point out here tat Dharmakirti not only
draws a clear distinction between perceiving and designating but also regards
designating as absolutely incompatible wit perceiving. His argument is tat no
verbalization can take place in te absence of conceptualization and that tere is
no conceptualization i perception. In te expression "abhilapasanisargaoga"
te word "yoga" (capable) is quite significant. It clearly states tat though it is
necessary tat tere cannot be verbalization witout conceptualization, the reverse
is not the case. That is, it is not necessary tat where there is conceptualization,
there must be verbalization. As Dharmottara writes, " We may also have
conceptualization, which altough not accompanied by corresponding words, is
capable of being accompanied, as for exanlple, the conceptualization of a baby
which has not been verbalized. "
15
Wherever tere is verbalization, there has to
be conceptualization, but whether te reverse is also te case can be a debatable
point. It would, terefore, be in order if we refect on the relation between
conceptualization ad verbalization in the context of Dharmaklrti ' s position.
Following Dharmottara, it can be maintained that by inserting the word
"yoga" DharmakIrti wants to point out tat it is not necessary .tat every case of
conceptualization should be accompanied by verbalization. Every concept is
capable of being verbalized, but, in fact, it may or may not be verbalized. This
means only that verbalization is not icompatible wit conceptualization in te
way in which it is incompatible wit perception, but it is not its necessary
accompanilent. However, if we keep in view a verse generally ascribed to Diaga
in which concept is regarded as having its "yoni" (source or place of origin) in
words,
16
then Dharmakirti ' s position can also be understood as signifing tat
tere are two modes of verbalization, namely, actual and potential . Though all
conceptualizations may not be associated with actual verbalization, potential
verbalization is necessarily present terein. Such a distinction between " surface
language" and "depth language" might have been intended by Dharmaklrti .
Accordingly, in every conceptualization tere must be necessarily depth language,
which means that there can be no ting witout te use of language. The use
of language may not be overt but covert.
Besides kalpanaporham, the other constituent of Dharmaklrti ' s definition is
"abhrantam. "
1 7
Perceptual knowledge is tat which is free from kalpant and at
te sanle tinle is abhrantam. Wile explaining te teml "abhrantam, " he mentions
diferent instances of error caused by color blidness, rapid motion, traveling in
a boat, mental sickness, and so on. Commenting on it, Dharmottara writes that
absence of kalpana and absence of bhranti (error) taken together, not separately,
constitute the defnition of perception. Dharmottara poits out tat tese four
diferent illustrations represent four different types of illusion. The cause of
color blindness is located in the sense organ (indriyagata) . The cause of rapid
32 Buddhist Epistemology
motion resulting in the cogmtlOn of a fiery circle is located in the object
(viagata) . The cause of illusion of moving trees is located in te exteral
circumstances that condition the perceiver (baJas 'raasthita) . Lastly, the cause
of hallucinatory illusions is located in the interal circumstances (adhatmagata)
like the mental state of the perceiver. All these causes, whether located in the
organ or in te object, whether exteral or iteral , affect the cognitive senseI S
and result i n illusory sensation. So, perception should be a sense cognition tat
is distinct fom these illusory sensations.
1
9 If we undertake a comparative analysis
of the definition of prtak$a given by Diiaga and DharmakIrti , three basic
problems come up for our consideration. They are:
1 . Is prtak$a a prmaza or samyag jiana (kowledge) ?
2. What is meant by kalpana? and
3 . I s the adjective "abhrantam" necessary?
IS PRTAK$ A PRMA?
Both Diimaga and Dharmaklrti , before giving a defition of prtaka,
point out that it is one of the two prmalJas. Of course, in Dharmaklrti we fd an
explicit a detailed analysis as to what is meant by pramalJya or samyakatva
(truth) , and no such detailed analysis is available in DiInaga as far as our
information goes. It is, however, incontvertble tat bot of tem rga praa
as prmalJ a. However, in the context of teir insistence on kalpanapohatvam
(nonconceptuality of prtaka) , it can really be asked whether they can justifably
regard prtak$a as a prmalJa. Santarakita20 defnes perception exactly on the
pattern of Dharmaklrti without using the word jianam or pTmalJam, and tere
Kmalasila, his commentator, raises a point as to why te word jianam is not
used in this defition. Does it in any way mean that Santarakita intends to
exclude prtak$a fom the sphere of jianam? To this KmalasIla responds by
saying that te wordjianam is not used in the definition because it is implied and
therefore not needed. However, it can certainly b inqu a t wheter prtaa,
as it is defined by tllese Buddhist thinkers, can be a prmalJa, and if it can be,
then in what sense? An answer to tis question depends on our understanding of
prmtza. If by prmala we mean just the presence of avismilvadakatam witout
tere being any necessary awareness or confrnlation of it (asandigdhatvam.) ,
prtak$a can be regarded as prmala because it is having avisarhvadakatva
with its object insofar as it is exclusively determined by the object (arthaniyata)
and is of te same form as te form of the object (arthasanlpya) . But if prmaza
is understood in terlS of avismivadakatvam and asardigdhatvam, tat is, a
cognition that is true and also indubitable, then, certainly, prtak$a canot be
regarded as a prmala because indubitabilit or certainty i s someting that is
dependent on confirmation or on the production of some invincible grounds.
This would mean that only a vavasayatmaka Uudgmental) or savikaiakajiana
(conceptual) can be a prmalJa, and since prtak$a by its very definition is
The Buddhist Theory of Perception 33
nonconceptal , i t camlot be put under the rbric of prmafa. But as we have
said, if indubitability is not to be regarded as an essential element of prmaf.a,
then the p
r
afatva of prtaka becomes a matter of aalytic consequence.
WT IS KPAA?
The other poit tat arises for consideration is understanding the word
"kalana" by Diaga and Dharmaklrti . Sice Diiaga uses the expression
"namajataiojana" to defie kalpana, whereas Dharakrti uses te expression
" abhilapasa1lsaraogaprtbhasaprtfi" for tis purpose, it has to be examined
wheter bot of tem are using the word kalpana in the same sense or in two
different senses. In other words, what needs to be explained is the meanig of
kaZpana according to Diiaga and DharmakIrti. When Diiaga defes kalpana
as " namajattdiyojana, " the question arises whether it is to be understood as
yojana (association) of nama, jati, and so on with thigs or association of nama
with jati, and so on. If it is understood in te former sense, then it goes against
Diiaga's basic position tat the ting (svalaka{za) is beyond all associations.
Evidently, therefore, it is to be understood only i the latter sense, 21 as has been
explicitly stated by Santarakita. 22 Understood in tis sense, tere does not seem
to be any difference of opinion between DiInaga' s and Dharmaklrti ' s
understanding of kalpana. Santarakita defnes kalpant as "abhilapiniprtfti" on
te pater of DharmakIi but takes elaborate pais to poit out that it does not
go against Diaga' s defnition given in the Naa MUkha. 2
3
IS TH AJECTI " ABHAM' NECESSAY?
The necessity of mentioning the characteristic of abhrantam has been
controversial right from the tinle of Diiaga. It is frst mentioned by AaIga, is
dropped by DiIaga, but is reintroduced by Dharmaklrti . It is again dropped by
some of his followers like Vrtadeva but finally reestablished by Dharmottara.
However, it must be noted tat Diiaga is not alone in not using te adjective
" abhrantam. " As Sataakita points out, several Buddhist tinkers regard ilusions
as purely mental facts having nothing to do with sense perception. These thiners
object to the inclusion of the adj ective " abhrantam" in the definition of
prtaka.
2
4
Wy Dilnaga dropped the characteristic "abhrantam" is not known for
certain, but Stcherbatsky25 has pointed out three considerations that seem
to be quite plausible. First of all , every illusory cognition has to be judgmental,
but no judgmentalization belongs to the sensuous part of the cognition.
Every judgment is the construct of the intellect and not the refex of the
senses . DiInaga maintains that pure sens ation does not contain any
judgment, neither the right one nor the wrong one, because it is nonconceptal .
It can ot contain any illusion at all , a hence there is no need to add the adjective
"abhrantam. " However, this doesn' t mean that perceptual cognition is not a
pramafla because for him pramafl a consists only in arthasanlpya or
avisGlivadakatva and not i n asamdigdhatva.
34 Buddhist Epistemology
Second, Diiaga rejects Vasubandhu' s defmition of prtak$a because it
could be given a realistic interpretation. For the same reason he resolves to drop
the characteristic of abhrantam because it could be interpreted as maintaiing a
position tat all conceptual lowledge is erroneous and that perceptual lowledge
alone is nonerroneous.
Third, Dinaga is aware of te ambiguit of te word "abhrantam, " and
to avoid confsion, he must have dropped it. Diiaga, of course, doesn' t deny
that there are illusions, but tey must be treated separately. Just as there are
inferential fallacies, hetvabhtsas, sililarly there can be prtak$abhtsas, but
these pratak$abhasas are due not to the senses but to the intellect only.
Dharmakirti disagrees with Diraga on this poit and reintroduces te
characteristic of abhrantam i te defiition. His main argument is tat we
must distinguish between sense illusion and mental illusion. For example,
when we mistake a rope for a snake, the illusion is purely mental and ceases
when it is dispelled. But when we see the moving moon i the clouds , this
illusion continues even if we are convinced that it is the clouds that are
moving.
While explaining Dharmakirti ' s contention, Dharmotara points out that
the purpose of using the two characteristics is not just to distinguish prtak$a
fom anumana but also to clear away wrong cognition. 26 The characteristic of
kalpanapoham would alone have been quite suficient for that but if abhrantam
were not added, erroneous cognitions, like that of a movig tree, tat are not
faulty mental constructs but are due to sensual aberrations would not be
distiguished fom genuine perception. Therefore, Dharmottara isists that the
introduction of abhrantam is perfectly i order.
In this context Dharmottara writes tat the term "abhrantam" should not
be understood in the sense of avismilvadakam because this would iply tat
perception alone is avisalilvadaka and not anumana because anumana doesn' t
represent svalak$afa. Had it been Ds in to maintai tat praka
alone is avisarvadaka and not anumana, the term "kalpanapoham" would
have been sufcient for tat because all kalpana wuld b va. But tat
dosn't see to be te case. For Dharmakrti both prtak$a and anumana are
avisamvadaka in the sense that bot are consistent with their respective objects,
namely, svalak$afa and samanalak$ala.
Santarakita and Kamalasila defend Dharmakirti and argue at length to
establish that illusions occur due to sensual aberrations also, and they are
not just to be explained as mental distortions. Ultimately, all illusions can be
traced to a defect in the sense organ concered, and if the illusions were
not due to this defect and were due to mental aberrations, tey would have
disappeared when te mental distorion is rectifed. Not only this, but te vivid
presentation of false objects i illusions canot be satisfactorily explained unless
they are regarded as sensual presentations. 27
Dharmottara explicitly mentions that there were some thinkers among
Buddhists themselves who maintain tat even these abnormal experiences. are to
The Buddhist Theory of Perception 35
be regarded as valid knowledge inasmuch as they satisf the pragmatic tests. He
writes, "There are some who maintain that te vision of a moving tree by an
observer traveling by ship and similar perceptions are right perceptions because
there is in this case .an underlying realit which is not a construction. Indeed a
man actig upon such a perception reaches something which is a tree; hence it is
supposed tat experience supports his perception. "28
Similarly, Santarakita also writes that because of a successfl outcome even
in erroneous cognition "some people" maintained that erroneous cognition should
also be regarded as valid. 29 Wile commenting on tis, KamalaSIla explains that
"some people, " in fct, means some Buddhist thinkers themselves. 3o
Bot Dharmottara and Santarakita point out that what constitutes validity is
not pragmatic fitess alone but that plus consistency (avisarhvadakam) wit reality.
So such presentations, as the light of the jewel for te jewel itself or of yellow
conch shell for a really white conch shell or of moving trees for trees that are
really stationary are not valid perceptions, though there is actual verifcation.
Mere verification and pragmatic satisfaction cannot, however, be accepted as the
test of validity; but consistency of presentation with reality is te criterion. What
is presented is the light of the jewel or the yellow conch shell and what is actually
attained is not the light of the jewel or yellow conch shell but something different.
In te mirage, too, te refacted light of the sun and the determinate experience
of water are presented.
Having discussed the divergent views of the Buddhist logicians from Diiaga
onwrd, we can while concluding restate our problem. Is not Diaga' s defition
of perception as ka/panapotham complete and sufcient by itself I seems tat
addition of the adjective " abhrantam" is not logically necessary. Probably,
Dharmaklrti thought that the term "kalpanapotham" is competent only to
distinguish prtaka from anumana but not from bhranta jfana. But here we
can question and perhaps reject Dharmaklrti ' s contention and uphold Diiaga' s
position on te following grounds.
Indeed, sense perception can be reckoned as a source of knowledge
(rmaza) only under the condition that the knowledge produced by a
sensation does not represent the sensory illusion. However, it seems quite
superfuous to mention " abhrantam" as characteristic of valid sense
perception, because, according to the classification of the system, sense
perception is a variety of valid, that is, nonillusive, cognition.
The' term "abhrantam" is almost a synonym of "kalpanapotham, " since
construction represents illusion when compared with pure sensation, the
genuine source of knowledge. So it becomes useless to use the synonym
in the definition.
A question may arise, How can we have erroneous cognition at te level of
perception? Perception being nirvikalpaka, we have just te sensations of the
object, and no judgment is formed at tat stage. Ony judgment can be right or
wrong. For example, "A" enters a room wit a friend of his, and there is a rope
lying in the room. Mistaking this rope for a snake, "A" cries, " Snake "
36 Buddhist Epistemology
The friend says, "No, it is merely a rope. " Thus, the perception of the same
object, rope, has resulted in two diferent mental constructs. The rope must have
given the same kind of sensation to bot of them, but the first one thought it was
a snake, and the second one tought it rightly to be a rope. So, mental construct
results in illusion, not bare perception. For distinguishing mental construct fom
perception Diiaga' s defnition is competent enough, and there is no need to add
the adjective " abhrlntam" to it.
Dharmakirti, on one hand, believes in sensual aberrations and, wit a view
to exclude these from the category of perception, includes te term "abhrlntam, "
but, on the other hand, he strongly advocates the theor of svata!-prlmlryavlda,
according to which all cognitions qua cognitions must be true. This fact Camot
be overlooked. At the level of sensation tere is no error. From tis it follows that
errors are due to noncognitive factors, that is, factors extraneous to sensation.
Hence, the addition of the adjective "abhrlntam" seems to be useless.
NAT OF PRT$
All the epistemological thiners in India are unous i rgagpr'
as a reliable and valid cognition, but there exists a wide difference of opinion
wit regard to its precise nature. Ordinarily grantig a basic distinction between
two types of cognition, nirikalaka (indeterminate) and savikalpaka (determinate) ,
it has been debated as to whether prtaka is to be regarded as nirvikalpaka
only or savikalpaka only or bot nirvikalpaka and savikalpaka. The Grammarian
philosophers and te laina thinkers take one extreme view tat all cognitions,
icluding perceptual ones, are savikalpaka since they are predicative in character
and also necessarily embodied i verbal propositions overt or covert. The Buddhist
position lies on the oter extreme, that prataka is always nirvikalaka and can
never be savikalpaka. Anllmlna alone is savikalpaka. Before we discuss the
Buddhist position in detail , it would be worthwhile if we briefy outline the
conficting non Buddhist positions on this issue.
Advata Veditic Vew
In the Advaita-Vedanta, particularly in its earlier phase, we seem to have a
position akin to the Buddhist one, which is arrived at on a slightly diferent basis.
According to it, the realit being of the nature of pure existence bereft of all
characterizations (sllddla or nirvise$a sanmltrm) , the perceptual cognition that
is in the form of presentation of an object must be nirvikalpaka. It is a cognition
of "tis " or " te existent" but not as determined by anyting. To detemline a
thing is to characterize it by this or tat qualit and is thus to distinguish it fom
other things having different qualities. All this is .the work of buddhi. When we
have a determinate cognition of an object on te basis of discrimination, the
object is detemlined by certain qualities and relations. Such a cognition, which is
ordinarily called savikalpaka prtaka, is not valid (mithl) because it is no
longer representative of the object, which in the ultimate analysis is one
undiferentiated consciousness (caitana) . The Neo-Advaita Vedanta, however,
The Buddhist Theory of Perception 37
accepts a distinction between nirikalpaka and savikalpaka prtaka from the
vavahtrika or practical standpoint, tough from the parmanhika standpoint
such a distihction is untenable. According to it, savikalpaka prtaka is the
kowledge of a ting as qualifed by an attribute. The essence of savikalpaka
prtaka is the determination of its object by way of predication. On the other
hand, nirikalpaka prataka is kowledge of the bare existence of a ting without
any predication. Strictly speakig, propositions lie " Tatvamasi " alone can be
regarded as genuine nirikalpaka prtaka. Through "jahat-ajahat-lakala"
such propositions are to be understood as conveying the siple identity between
"tat" and " tvam. " Such propositions, technically known as akhmitanha vaka,
are nonpredicative in character. For example, when we say "This is that man, "
we do not predicate "that man" to "this. " It is not a case that we relate "this
man" with "that man" but sily assert the identity between the two. 3!
Gramarian's Vew
Granmarians go to the other extreme, and in fuis Jainas join them in
maintaining that all perceptions are savikalpaka because they are necessarily
predicative in character and verbalized in form. Not only perception but every
cognition, perceptual or nonperceptual , has to be both propositional as well as
sentential . In other words, tey maintain that te epistemic content and the linguistic
form are always inseparable, and all propositions are invariably linguistic in
form. There is no pure thinking, but "speech ting. " In te words of Bharrhari,
a Grammarian philosopher, "There is no thought-content without a language
form and always thought-content is inextricably interspersed with language form.
If the unfailing form of language had been lost to the thought-content, thought
itself would have been lost in oblivion. Thought has a form because it is formed
in language. "32
From this it follows that all cognitions, including perceptual ones, are at
once both judgmental and verbal . A corollary of this would be that perception
can never be nirvikalaka but should always be savikalpaka.
Nyaya-Misa-Sarya-Visitadvaita Vedanta Views
In te schools of Nyaya, Mlamsa, Sar, and Visitadvaita Vedanta,
the preceding two extreme views are reconciled, and a middle position is put
fort. It is held i these schools that any perception, nirvikalpaka or savikalpaka,
is a direct cognition of a real individual that is a unity of universal and particular.
The distinction between nirvikalaka and savikalpaka is due not to any difference
in the contents of perception but to the wy in which the same contents of experience
may be ordered and arranged.
The Samkya tinkers perhaps initially advocated a distinction between two
stages in perceptal cognition, namely, nirikalpaka and savikalpaka, and used
the term "alocana" (Gestalt view) for nirvikalpaka prtaka. The first
apprehension of an object is in the form of undifrentiated sensum, and te
manas is salikalpaka, tat is, difrentiates te undifferentiated sensul presented
by the senses. While commenting on te Sarhkha Karika Vacaspatinlisra, in his
38 Buddhist Epistemology
StThka Tattva Kumudl, introduces a fll-fedged distinction between the two
kinds of perception. 33
Kmarila of the Pirva-Mlmamsa school seems to follow the Sarkhya line.
For him the nirvikalpaka prtak$a is the first moment of cognition in which
there is an undifferentiated apprehension of the obj ect as one individual whole of
generic and specific attributes (sammugdhavastllnttr) . Here tere is no definite
cognition of it as this or that object (vastu vise$a) . Nrikalaka prtak$a thus
resembles children' s and dumb person' s perceptions. 34 It is nonverbal in nature
because tey lack te capacit to communicate. However, Kunlarila' s position
can also be understood in a diferent manner, if we take into consideration te
following verse and Partasarati' s commentary thereon.
Na vife$o na stmtnam tatnim anubhayate
Taortdhirbhatastu vaktrewivasi$ate.
(In that apprehension neither the particular nor te universal is experienced.
What is experienced is ony the individual obj ect tat underlies the two) . 35
That is, he regards nirvikalpaka prtak$a as a sort of inchoate cognition i
which neither the particular nor te generalized aspect is cognized; nor are they
distinguished. Whatever is te interpretation of Kumarila' s position, for him it is
indeterminate or nonconceptual and therefore nonverbal . The second stage is
tat of savikalpaka prtak$a, which is a determinate cognition in which the
object is cognized by the mind along with its characterizations.
Like te Saya and MIamsa tikers, Nyaya-Vaiseika tinkers also
accept both stages of perception, and most of them regard them as equally valid.
From Vacaspati onward we find a clear-cut distinction between the two stages of
prtaka. On te distinction between nirikalpaka and savikalpaka Jayanta writes,
"A universal, a substance, a relation or a qualit whatever essence of an object i s
grasped by savikalpaka is equally grasped by nirikalpaka. The remembrance of
a meaning-relation of te object wit a certain word is te only relation tat
marks out savikalpaka fom nirikalpaka. So far as the obj ect is concered there
is not the least difference between the two. "36 From tis statement the only thing
we can distinctly make out is that for some Naiyayikas, the distinction between
nirvikalpaka and savikalaka is only in verbalit. The savikalpaka is verbalized
and predicational , whereas nirikalaka is nonverbalized and nonpredicational .
So in te nirvikalpaka we apprehend the object along wit all its relations and
characterizations, but we do not interrelate them (vaifi$.ttnavagthi) , and in
savikalpaka we do so. In the Navya-Nyaya, te distinction between the two is
more sharply drawn as compared to Jayanta. However, the consequence of such
a sharpness has been some sort of rethinking on the natre of prtak$a in the
Navya-Nyaya. Visvanata, in te Bh$t-Pariccheda and Siddhinta-Muktvali , is
unwilling to accept nirvikalpaka within the fold of prtak$a since it is a kind of
prejudgmental cognition. 37
In the Visitadvaita Vedanta, even though te distinction between nirikalpaka
and savikalpaka is accepted, the two are regarded as diferent quantitatively.
Both are predicative and relational in character, but whereas te frst one is vague
Te Buddist Teory of Perception 39
and indefinite, te second one is vivid and defnite. The frst is the itial awreness
(rthama pifrha{za) of the object. It is about te total object in which the
individual as' well as the universal aspects of the object are apprehended in teir
interrelationship, but there is no complete certainty. In the second or successive
apprehensions the judgment becomes decisive and is therefore known as
savikalpak. Ram: uja understands savikalpaka not i terms of judgmentaliation
but in terms of more decisiveness or certainty.
Buddhst Vew
A against Grammarians, on one hand, and Naiyayika, MImamsaka, and
SaIl ya thinkers, on the other, Buddhist thinkers insist on perception being
nirikalpaka. A clear-cut distinction between nirikalaka jitna and savikalpaka
jitna seems to have been introduced for the first time by Dinaga. According to
him, the diference between nirikalpaka and savikalaka is not just quantitative,
a difference in vividness and sequence, but is fndamental and qualitative. The
nirikalpaka jitna is a cognition of an object-qua-object that can be ony in the
form of pure awareness and that can never be in the form of awareness in terms of
thought. The reason is that our thought can never grasp the object-qua-object
that is in the form of svalaksala because our thought is always restricted to
stmtnya lakala. It is thus clear that the two tpes of knowledge are quite different
not only in their natre but also in the spheres of their operation.
While defining prtak, thikers fom Dinnaga onward always insist on its
being kalpanapoha. Kalpant stands for both thought and speech, and, therefore,
the phrase "kalpanapoham" iplies two things : (1) i is nonconceptual , and
(2) it is nonverbal .
Buddhists start with a dichotomous distinction between a kowledge that is
perceptal and a knowledge that is nonperceptual , that is, conceptual . From this
it follows that perceptal cognition cannot be conceptal cognition, and conceptual
cognition cannot be perceptal cognition. The nonconceptality of perception
means its feedom fom all mental conceptions. In other words, all conceptual
contents are mind-given, and mental impositions, whereas perception is given by
the object and not at all by the mi. Dil naga38 quotes an Abhidhamma treatise
in his support where it is stated that one who has ability to perceive, perceives
something blue (NUam vitnttl) , but it is not conceived tat "this is blue" (NUamiti
vijantti) . The expression "NUam viantti" implies that one has an immediate
awareness of the object itself. On the other hand, "Nllamiti vianti " implies that
one forms a judgment based on perception by associating a name with the object
perceived. Thus, the preceding passage expresses the idea that perception is free
from conceptual construction. In other words, in respect of an object te perceiver
has a sensation of the object but does not possess any conceptual awareness of it.
There are two things we mean when we say that perception is nirikalpaka:
1 . (a) In all perceptions we never perceive the universal or te universal qualifing the
40 Buddhist Epistemology
individual. Thus, we always perceive an individual, the svalama, and nothing
more than that because individual is absolutely beref of all universals. At the level
of svalaala there are no qualifications, and Buddhists are quite vehement in
their rejection of dhamza-dhannl bheda (substance-attribute relationship) because
in perception tere is no apprehension of qualities and thus there is no determination
and so on. For example, "X" is an object, ad if "X" is indeterminate ten "P" as
te kowledge of "X" should also be indeterminate because only then can
knowledge be in agreement with its object. If "P" is determinate, then it would
mean that it is not given by the object alone because the object has no determination
and cannot cause any determination. Consequently, it will be in discrepancy with
the object. If the perceived object is indeterminate, its knowledge has also to be
indeterminate.
Every svalaala that alone is te object of perceptual cognition is instataneous
and does not endure for te next moment. Ever svalaala, being momentay,
canot be perceived twice, and terefore every perception is diferent from another
perception. Like svalaala, every perception is unique (svatovtvartaka) .
(b) There is another sense in which perception is nonconceptal. The conceptual
cognition (saviklpak jitna) is always vaifi.ttvagahi jiana (predicative cognition)
and smizsargavagthi jitna (relational cognition). In other words, it has to be
propositional. Every proposition is a mental construct, and without mental
construction there cannot be any formulation of a proposition. Thus, a perceptual
cognition is nonpropositional. Though in itself it is nonpropositional, on the basis
of it one can formulate a proposition. In this sense it . can be regarded as
prepropositional. But the moment prepropositional perceptual cognition is
transformed into a proposition, that is, conceptualized, it ceases to be perception.
To be conceptual is to be nonperceptual.
2. There has been a broad agreement among Indian epistemological thinkers that no
determinate knowledge is possible without a language. Buddhists also fall in line
with this. Though al determinate knowledge is linguistic, not all knowledge is
linguistic. Perceptual knowledge that is indeterminate is nonlinguistic because it
does not require the help of language and also because it is beyond the reach of
language. The nonverbality of nirviklpaka prtaka is put forth by Nyaya
Mimamsa thinkers, perhaps on the frst ground rather than on the second, but for
Buddhists the second ground is more signifcant than the first one. Prtaa is
anabhiltpa (inexpressible in language) not only because it does not require the
help of language but mainly because it is determined by the object (svalakaza) ,
which is beyond the reach of language.
Buddhists put forth vigorous effort to defend teir position vis-a-vis te non
Buddhist ones. We may here present in brief the Buddhist arguments for the
rejection of the non-Buddhist positions, which can be broadly represented into
tree, namely, the Nyaya-Vaiseika, the Jaina, and te MImamsaka.
Refutation of Nyaya Position
In the Nyaya tradition from Vacaspatimisra onward we have a clear-cut
distinction between two stages of perception and hence of two types of perceptual
cognition. The Buddhists accept nirikalpaka alone as prtaka, and therefore
they would deny that there are two stages or types of prtaka in the form of
The Buddhist Theory of Perception 41
nirikaZpaka and savikaZpaka. The main argument put forth by the Buddhists for
nonacceptance of prtak$a as savikaZpaka is that all determinate perception is
capable of heing verbalized, and if perception is to be defined in temS of
kaZpanapoha or in terms of artha-pratta (being caused by the object) , then it
can never be associated wit verbal expression. A cognition that is in the form of
apprehension of an object (prtibhasa) can never be associated with verbal
expression because neiter are there words present in the object, nor are the
words identical with the object. 39 If it were so, a person ignorant of the meaning
of the words would also be capable of communicating through words like the
person who knows their meaning. In other words, to cognize is one ting, and to
designate is another. The cognizing fnction is not identical with the designating
fnction, nor is te designating fnction inseparably associated with the cognizing
fnction. Therefore, a cognition produced by te object should present ony te
object and not the word, which is wrongly supposed to be denotative of the
object. From this it follows that a determinate knowledge that is a mental
construction is erroneously held to be a true perception. Detenninate knowledge
presents an object as associated with a word (abhilapa) , although in realit the
object is independent of any association with a word. Determinate knowledge is
caused by the past mental ilpression (vikaZpa-vasana) , and it grasps an uncertain
(aniyata) object, not a fxed one. It difers with every individual because it is not
a real object and is caused by the past mental impression. The fnction of mental
impression is always concealed in a determinate knowledge, but since that
knowledge follows in the wake of perception, it usurps te fnction of te latter,
and because of this detenninate knowledge is mistakenly supposed to be perceptal.
The other argument put forth by Buddhists is that the detenninate kowledge
is always i predicative form in which someting is predicated to something
(vi$!avaiavagahi jfana) . Thus, it is always predicative of a qualifer-qualifed
relationship. According to Buddhists, first , there is no qualifier-qualified
relationship existing between the objects (svaZak$a7las) , and second, even if such
a relation is objective, one single perceptal cognition cannot apprehend two
objects with teir interrelation because it is essentially simple in character and
cannot be complex.
Reftation of Kuarila's Position
Santarakita40 undertakes an elaborate examination of Kumarila' s view that
tere are tw stages, namely, nirikaZpaka and savikaZpaka, necessarily involved
in perceptual cognition. His main objection is regarding perception as detemlinate
cognition. If in a perceptal knowledge both the qualifer and the qualified, the
individual and the universal , are apprehended, then the question is, How can tis
be possible in a single act of perception? Even granting that universal is
perceptally apprehended, when universal is apprehended, individual is not
apprehended, and when individual is apprehended, universal is not apprehended.
One and te same cognition canot apprehend bot. Then how can a determinate
42 Buddhist Epistemology
perceptual knowledge be possible? Further, if it is maintained that in the
indeterminate cognition itself both the individual and universal are apprehended,
then what is lef to be apprehended at te determinate level? As a matter of fact,
universals and so on do not exist eiter as diferent or nondiferent from the
individuals. Nor can they be bot different and identical .
Refutation of the Jaina Position:
The position that every cognition, let alone a perceptal one, has to be
determinate is based on the argument that every object is twofold in character,
namely, the individual and the universal, and one cannot kow an object without
at the same time cognizing both its aspects. If an object is not apprehended along
with its diferential character, then there will be no difference between one
cognition and another cognition, and te cognition of pen will be the same as the
cognition of paper. The assertion that a cognition apprehends a qualifed thing
and yet is not conceptal will be self-contradiction. Thus, te Jaina argument is
that to cognize is to cogize a thing with its differential character, which means
it has to be a determinate cognition.
To this Santarakita' s41 reply is that it is one thing to apprehend a qualifed
thing, and it is an altogether different thig to say that a thing is associated with
qualifications. A thig can never be associated with qualifcations because by its
very nature everythig is a unique particular, but every particular tig is different
fom anoter tg, and this fact of being "diferent fom" enables us to distiguish
one perception fom the other perception. Every perception is a negation of all
things other than itself, and this accounts for the differentiation fom the rest. In
fact, all perceptal cognitions are unique and distinct just as their objects are
unique and distinct, but the awareness of their distiction is conceptal , and this
awareness, which is conceptual, is certainly not perceptal . The perceptual leads
to conceptual, but perceptual itself is not conceptual . From the preceding analysis
Buddhists conclude that all cognitions concerning svalakms have to be
compatible with the nature of the svalakazas. The svalakmas are devoid of all
characterizations, and therefore their perceptual cognition also has to be devoid
of all characterizations. .
On the basis of the foregoing analysis of te Buddhist and non-Buddhist
positions on the natre of prtaka, we can restate the Buddhist position
consistently and unambiguously put fort by Vasubandhu, Dinaga, Dharmaklrti,
and so on as follows :
1 . Patala can never be conceptal or deterinate (savik/pak) mainly because it is
exClusively determined by the object, which is free from all determinations and
characterizations.
2. Nor can perceptual cognition be both indeterminate as well a s determinate or perceptual
as well as conceptua, mainly because perceptual and conceptual are mutally exClusive.
A perceptual can never be conceptual, and a conceptual can never be perceptal .
This does not mean tat a perceptual cognition has nothing to do with a conceptual
The Buddhist Theor of Perception 43
one. Though in itself it is nonconceptual, it may give rise to conceptual cognition and
thus may become pretonceptual. But a cognition remains perceptual only insofar as
it is nontonceptual and ceases to be so the moment it gives rise to conceptualization.
TES OF PRTS
All Buddhist thiers of te Vifanavada tradition fom Vasubandhu onward
are unanious in insisting tatprtaa is nirvikalpaka and hence kalanapoha.
Thus, the differential character of every perceptual cognition is to be free from
all conceptual constructions. In this respect they are all alike, and they are not to
be classified under different heads. Howevr, as Di naga42 points out, they can
be classified if such a classifcation is asked for. Though we have a rough
classifcation of diferent types of prataa in the Prmaza Samuccaa, a precise
and clear classification is available only in Dharmaki ' s works. Perhaps, Dilnaga
is more concered about givig a precise definition of prtaa rather than
about the different types of prtak$a. For him classifcation of prtak$a is
logically not signifcant. However, tere was a prevlent practice in philosophical
circles to discuss types of prtak$a. Di naga refers to it, but Dharmaklrti feels
a need to expound the Buddhist position on te classifcation of prtak$a i
clear-cut terms.
Diaga's Classifcation of Prtya
Having stated tat all perceptal cognitions are alike so far as they are fee
from conceptual constructions, Dilnaga points out that tey can be classified
into indrya prtak$a (sense perception) , mlnasa prtaa (mental perception) ,
svasarvedana pratak$a (selfcognition) , and yogi prtak$a (mystic perception) .
1 . Idiya pratya. So far as the indriya prtak$a is concered, it is in
the form of sensations caused by an exteral object. Since there are five kinds of
cognitive senses, we can talk of fve types of indriya pratak$a.
2. Maasa pr8ty8.43 Besides indriya prtak$a tere is also manasa
prtak$a, which consists of te awareness of indriya pratak$a.
3 . SV8savedaa prtya. The third type of prtak$a is the immediate
experience of bot indriya pratak$a and mlnasa prtaa.
The concept of svasOvedana has been one of the most significant
contributions of Dilnaga, and its implications i te context of prlma1yavada
have been worked out while discussing sanlpa. The scope of svasGlivedana has
been enlarged by Dinaga to include the awareness of conceptual constructions
also. In his own words, "Even conceptual construction when it is brought to
internal awareness is to be regarded as a type of prtyak$a (kalpanapi
svasarhvittavi$tl) . "44 I t i s interesting t o note here tat, accordig t o Dilnaga,
. each cognition has a tofold appearance: te appearance of an object (arhabhasa)
and that of itself (svabhasa) . As such, cognition cognizes itself while cognizing
an object. The cogniing of an object through kalpan i a, notprtaa.
44 Buddhist Epistemology
But whether -it is anumana or prtaka, the essential nature of cognition is te
same; that is, it is self-cognized (sva-prkasaka) .
4. Yg prtya$a. Besides the previous three, the fourth type of prtaka
accepted by Diilaga is yogi prtaka. The intuitions apprehended by a yogi are
also nonconceptual and hence are to be placed under prtaka. Here Dilnaga
draws a distinction between a yogijfana and agamajfana. The cognition derived
from the agama (scripture) is kalpana, but the yogi jfana is not SO. 45
Dharmakirti's Classifcation of Pratyaka
When we come to Dharmaklrti, we fd tat te problem of tes of prtaa
seems to have been crystallized. Both i te N{a Bindu and in the PrmalJa
Varika we have clear statements that prtaka is of four types, namely, indriya
prtaka, manasa prtaka, svasarvedana prtaka, and yogi prtaka.
1 . Idiya prataa. So far as the indriya prtaka is concered, i is te
most primary and at te same tie te most pervasive tpe of prtaka. Such a
kowledge consists in presentation of an object to consciousness trough the
medium of senses. The cognitive senses being fve i number, it can frher be
classifed under fve heads. This fvefold classifcation is based on te diferent
cognitive senses, which serve as a medium. The cognitive senses are only a
medium, not an agent. Their fction consists only in creating a sort of link
between the subjective consciousness and te objective reality lying outside. This
fnction is over when te object is presented to consciousness and does not
continue thereafer. Since in te Buddhist tradition perception has been defined
as nonverbal , a question arises as to wheter tere can be anything lie auditory
perception. The question is based on an assumption that an auditory perception
that cognizes word cannot be fee from word association, and if wordassociation
is kalpana and not prtaa, how can auditory perception be a genuine perception?
Moreover, like light and consciousness, a "word" has a dual fnction of revealing
itself and revealing its object. Because of tis also an auditory perception must be
associated with verbal expression. The answer given by Dharmottara46 and
Santarakita47 is that when a word is cognized, it is not the same word that is
expressive of some concept. It is cognized only as a sound, which is svalakalJa
and which does not have the previously referred- to double character. Even if no
distinction is drawn between a soud ad a word, te object of auditory perception
is said to be havig te twofold character. In auditory perception we are concered
only wit its first character, namely, sound, and ony in a conceptal cognition
are we concered wit its second character, namely, meaning. 48
2. Manasa pratyak$a (Mano vifia). The second type of prtyaka
accepted by Dharmaklrti is manasa prtaka. It is a prtak$a tat inmediately
follows the indra prtaka. It is, in fact, te element of attention when an
indriya prtaka arises. That is why Dharmaklrti in the N{a Bindu
49
defnes it
as "mental sensation which follows sense perception, which is its inmediately
precedig homogeneous cause. " The manas prtak$a is generated by te indriya
prtak$a, which acts as its immediately preceding homogeneous cause. 50 If
The Buddhist Theory of Perception 45
manasa prtaka is to be regarded as different fom, and as an effect of, indriya
prtaka, then a question arises as to what its object is. Does it grasp the same
object that is already grasped by the concerned i
n
driya prtaka? If it does, then
it will not be a valid Gognition because it will no longer be "ajfatarthaprkasa"
(awareness of an object not hitherto known) . On the other hand, if it grasps what
is altogether ungrasped by the concered indriya prataka, then any .manasa
prtaa can be related to any indriya prtaka, or even te blid would perceive
color. 5
1
To this te reply given by Dharmaklrti and Dharottara is that te objects
of the two are different, and yet the two perceptions are interconnected insofar as
the indriya prtaka along wit its object gives rise to the manasa prtaka and
its corresponding object at te immediately succeeding moment. 52 This complex
phenomenon can be explained with te help of te diagram shown in Figure 2. 1 .
Figure 2. 1
Process of Miasa Ptyaa
Note: 01 is indriya prtaka viaya; C
1
is indriya prtaka;
0z is manasa prtaka viaa; Cz is manasa prtaka;
C
1
is samanantar prataya of Cz ; 0z is ilambana prtaya of Cz.
The need for te acceptce of manasa prta over and above indriya
prtaka can be questioned, and justification of its acceptance can be asked for.
To tis Dharmottara replies tat it is a necessary postulate of the system, and no
proof need be adduced for its postlation. 53 Stcherbatsky, 54 however, gives some
justifcation, which is not very unconvincing. He opines that after having made a
radical distinction between the role of indriya and kalpana, Buddhist thinkers are
in need of some connecting link beteen the two insofar as prtaka has to lead
to kalpana. This connecting l is established by postlating manasa prtaka.
There is some basis i Dharmottara for the suggestion given by Stcherbatsky.
Dharmottara argues that manasa prtaka has to be postlated as a l between
indriya prtaka a kalpana because indriya prtaka, being momentary, canot
directly give rise to conceptalization unless and until it is first attended to in
consciousness, and if manasa prtaka is not accepted, not only will tere be a
difficulty in accounting for a transition from indriya prtaka to kalpana, but
there will also be a faw of indriya prtaka beig invalid because it will then
not have generative eficiency (arthakriyakaritva) . The manasa prtaka, which
is sensory on one side and mental on the other, has te psychological necessity in
46 Buddhist Epistemology
order to link to heterogeneous types of knowledge, namely, perception and
conception.
In fact, the acceptance of manasa prtaka in the DiInaga-Dharmakirti' s
tradition can be traced back to the Abhidhamma literature, where we have a clear
acceptance of manovifana and manovinana dhitu in te list of seventy-five
categories ito which the whole sphere of kowledge and reality is reduced. In
the Nyaya Bindu TiM Tippa(li
55
it is stated that the conception ofmanovifana as
a tpe of prtaka was a necessar deduction from the import of a scriptural text
that declares, "Color is cognised, 0 mons, by twofold cognition, the sense
perception ad the mental perception iduced by it. " Another question may demand
our attention concering the temporal location of indriya prtaka and manasa
prataka. It can be asked whether some other indriya prtaka can operate
even at te time when we are having manasa prtaka in the second moment.
Both Dharmakirti and Dharmottara reply to this view in the negative. They are of
the opinion that if the sense organ were to operate even at a time when we are
having manasa prtaka, there would be no indriya prtaka at all . 56 This
point has been frther explaied and elaborated in the Tppwzi like this : " If we
assume that in the second moment the outer sense-organ is engaged just as it is
engaged in te frst moment, ten its fnction will also be the same, i . e. , it will
make the object present i our ken. Why indeed should then sense-perception
not arise in the second moment also; why should not both the moments receive
the same name of sense-perception ?"
57
Durvek's58 explanation of Dharmottara's
statement is as follows : "The first moment of the cognition series is regarded as
dependent upon the sense organ because it is ivariably concomitant with the
presence ad absence of the sense organ. If even mental perception were to have
invariable concomitance with the presence and absence of the sense organ, this
could not be given as a reason to establish a cognition as sense perception, that
is, it canot be called as sense perception on the ground tat i depends on the
sense organ. So, even the frst moment could not be estblished as sense perception
on the ground that it has invariable concomitance with the presence and absence
of the sense organ. "
It is significant here to point out tat te concept of manasa prtaka
advocated by DiInaga and Dharmakrti is altogether diferent fom the concept
of manasa prtaka advocated by Nyaya-Vae)i tr. I te Nyay-Vaeika
philosophy manasa prtaka stands for the perception of the qualities of the
self, but in the Buddhist tradition it stands for te mental perception caused by
the preceding sense perception. Though te object of manasa prtaka in the
Buddhisttradition is an interal one, nevertheless, it is caused jointly by the
external object and its sense perception. Nyaya-Vaiseika tinkers regard mind
as an antarendriya (iteral sense) , but for Buddhists of the Diaga-Dharmakirti
tradition mind is not an indriya; mind here stands for the consciousness (dtta)
or, strictly speakig, a specific moment of consciousness.
3. Svasamvedana. The third type ofprtak$a distinguished by Dharmakirti
is svasamvedana or atmasalivedana. He defines it as "Sarvam cittacaittanam
The Buddhist Theory of Perception
47
tasarhvedanam" ; that is, all consciousness, cognition (cita) , and mental states
(caita) are self-cognizing. Like Dinaga, DharmakIrti insists tat diferential
character of consciousness as opposed to matter is its self-awareness. Matter is
always to be known. trough consciousness, but consciousness can be known by
itself only. This teory is known as svayan'lprakdavada, te self-luminosity theory
of consciousness, which is a distinct contribution of Diiaga to epistemology.
The doctrine that every consciousness is also self-consciousness has been a
fndamental tenet of the DiIaga-Dharaklrti tradition. Ever cognition of the
object-external or interal is at the same time a cognition of tat cognition. Thus,
a cognition can also be understood as an awareness of awareness . To explain the
theory of svayam-prakasatva of consciousness; the analogy of a lamp is often put
forth. A a lamp illumines the objects in its surrounding and its own self at the
sanle tie, not being dependent on any extraneous fctor for its own illumination,
just so is consciousness self-luminous. ConIDlenting on Dharmaklrti' s definition
of svasali!vedana, Dharmottara writes, "There is no mental phenomenon
whatsoever which would be unconscious of its own existence. " Further, he writes
that such a self-awareness is not a construction nor an illusion, and hence it is a
tpe of prtak$a.
59
DhanlakIrti6 proves the self-luminosity of consciousness as follows. First,
he points out tat pleasure, pain, and so on are mental states, but tey are of the
natre of consciousness and they cannot be cognized by other cognitions. From
this it naturally follows tat all cognitions and mental states are self-cognized.
Dharnlottara explains this point wit the help of the following example. When a
patch of color is apprehended, we at the same time feel something interally in
the shape of some emotions. It is not possible to maintain tat a patch of color is
felt as being itself the pleasure it afords us, because we do not say, "This patch
of blue color has itself the form of pleasure. " We do not regard blue and pleasure
as identical, nor do we feel tem as te same. Therefore, we really are experiencing
pleasure as someting quite different fom the object blue, as somedling tat is
not equivalent to blue, and this is no doubt knowledge. Therefore, we do experience
our own knowledge. Self-consciousness is essentially a case of knowledge; it
makes present to us our own-self.
The self-luminosity of consciousness does not in any way mean bifrcation
of consciousness into a subj ect and an object, which would be absurd .
Consciousness and selfconsciousness are, therefore, interchangeable terms. In
Buddhist writings the word " cita" stands for consciousness in general , but the
word " caita" stands for feelings and so on, which are classed apart fom cognition
on the ground that feelings do not contain "intentionality" insofar as they are
purely inward in reference. Barring this difference, they are all conscious states,
. the conscient character being common to cognition and feeling alike.
How one and the same cognition can be both the subject as well as the object
has been aswered by DharlakIti by saying tat the cognition directly experiences
the fm of the exteral object refected in it and not te exteral object itself; so
48 Buddhi st Epistemology
only metaphorically is it said that cognition grasps the exteral obj ect, which, in
fact, simply causes te refection.
61
4. Ygjiia. Besides tese precedig tree tpes of perception Dharmaki,
following Diiaga, accepts yogi jfana as the fourth kind of perception.
Dharmaklrti defines yogi jfana as intuition of a mystic tat is produced fom
the subculminational state of deep meditation on transcendental reality. Buddhists,
like other mystics, believe i the fact of intuitive realizations, which are available
to some gifed persons like saints. In fact, the acceptance of yogi prtaka is a
necessary prerequisite of the acceptance of the sarajfata (omniscience) of te
Buddha. The Buddhistic way of noble life consists of three broad stages, namely,
prjfa, tUa, and samadhi. The experiences acquired in the state of samadhi are
perceptual ones because they are vivid, presentational , and direct. Knowledge of
the four Noble Truths and so on is an example of such a tpe of prtaka. I is
nonerroneous and nonconceptal contemplation on te real .
In order to explain te phrase "bhrtarthabhavana prkaraparantajam, "
Dharmaklrti describes the three stages of contemplation as follows.
First, he states tat yogi prtaka is generated by deep contemplation, and
it is vivid and fee from conceptual construction; aferward, he explains each of
tese qualifing terms in anticipation of te possible objections that might be
urged against tem. I migt be askd how te cogntion generated by contemplation
can be vivid. Dharmalrti proves this point by drawig our atention to the fact
tat the persons possessed of fear, sorrow, and passion vividly see nonexistent
things as real because of the repeated tought of those thigs. Again, someone
might ask, " It is conceded tat te cognition generated by contemplation is vivid,
but how can one accept it to be free fom conception? " Dharmaklrti answers that .
the cognition in question has, indeed, vividness, and just for tis reason it ceases
to be a conceptual construction. Conceptual constructions are never vivid.
Dharnlaklrti again observes that all cognitions born bf a deep contemplation on
either reality or unreality are vivid and nonconceptal ; but te cognition born of
a contemplation on reality is valid, while the cognition bor of a contemplation
on unrealit is invalid because te former alone is in harmony with successfl
activity or is efficient to lead to te attainent of te object pointed out by it.
This, in short, is how the cognition bor of a deep contemplation on reality is
vivid, nonconceptual, and eficient to generate the purposive action, and hence it
is a case of perception.
62
Chapter 3
Te Buddhst Teor of Ierence
IERNCE (AA) AS PRA
In diferent systems of epistemology diferent nubers of prmtlas are accepted,
and with te exception of te Carvaka system, te acceptance of anllmtna as a
prmt{za has been unanimous. Apart from prtaka Buddists regard anllmtna
as prmt{za. Generally, a distinction is drawn betWeen two kinds of knowledge,
nanlely, prataka (inmediate) and paroka (mediate) . Anlllntna, as opposed to
prtaka, is a mode that can provide only mediate (aroka) knowledge.
Anllmtna, it is inlportant to note, is at once a mode of knowing and a way of
reasoning. Thus, it has an epistemic as well as logical aspect, both of which are
inseparably coalesced into one. In fact, the former aspect is te basis for the
latter, and tat is why anllmtna is generally known as a mode of knowing.
Development of the Theory of Anumina in the Buddhist Tadition
Among the different systems of logic that developed i India, the Buddhist
system is of significance comparable to that of the Nyaya. Though systematic,
logical analysis seems to have begun wit Vasubandhu, DiTaga and Dharmaklrti
developed a fllfedged theory of anllmtna as also that of prtaka and apoha.
The development of logic in the Buddhist system and, for that matter, in the
entire Indian thought-has been within the epistemological context and therefore
began with the theory of prtaka. Some people are of the opinion that the
science of logic (Nytya ststr) developed in India out of a science of dialectics
and debate ( Vtda ststr) , but it seems more probable that it was not so much the
dialectics as the art of conducting philosophical debates in public ( Vtda vidhl
that was the ground for the subsequent rise of the science of logic, a theory of
anumina, more specifcally, the parirhtnwnina. Sice the techique of anumana
50 Buddhist Epistemology
was employed as one of te modes of acquiring kowledge and providig proofs
or grounds for the truth of tat kowledge, the science of logic was also kown as
the science of reason (hetu vidya or prmara sastr) .
The Buddhist teory of anumana has been developed in te context of the
theory of twofold prmaras, namely, prtaa and anumana, which, i tum, is
based on the view tat there are two types of object of kowledge (rmeas) . A
stated earlier, te Buddhists put forth a process view of reality as opposed to the
substantival view of reality advocated by the oter schools of thought in India.
Such a real beig in te form of discrete and momentary existence is given only
i our sense-experience, which is pure sensation devoid of all conceptualizations.
But for empirical purposes tey accept another view of realit (tough, of a
lower order) i which different tought categories i the form of substance,
qualities, motion, class character, etc .. and so on are hypostatized and imposed
upon the reality. The entire system of logic is ' interwoven around such mental
constructions and teir interrelations. However, it also aims at explaining te
relation between te dynamic realit and the static construction of thought.
Though te word "anumana" occurs i te Pitakas, it does not seem to
occur i te sense of a science of logic. In theAnumana Satta
l
a principle is laid
down that one should not do unto others what one doesn' t wish others to do unto
oneself. This is said to be based on the generalization that what is dislied by
oneself is disliked by others as well . This kowledge is said to be inferential .
Here one should infer (svartMnumana) as follows : "A evil person who is swayed
by evil thoughts is disagreeable and disliked by me; now if I were to be evil and
swayed by evil thoughts I, too, will be disagreeable and disliked by oters. " If we
analyze tis anumana, it, in fact, consists of two anumanas, namely:
1 . I dislike a evil person, X is like me (as a person), Terefore, X (probably) dislikes
an evil person.
2. X dislikes an evil person (conclusion of ( 1 ) , I a a evil person, Terefore, X
dislikes me.
This apart, we also meet with the expression "anvaa jianam, "2 meaning
inductive kowledge in all the Nikaas as well as i te Abidhamma. By tis is
meant te inferential kowledge of te inductive patern, which is based on a
belief in te uniformity of nature and universalit of causal relationships. In te
Sutta Nikaa we come across several statements expressing causal relations such
as "wit the arising of birth there is arising of decay and deat, and with te
cessation of birth there is cessation of decay and death and so on. "3 Knowing
such causal relations is called "the kowledge of phenomena" (dhammajianam) .
It is frther stated, "This constitutes the kowledge of phenomena; by seeing,
experiencing, acquirig kowledge before long and delving into these phenomena,
he draws an inference (naam net!) with regard to te past and fture as follows :
All tose recluses and brahmaras who thoroughly understood the nature of decay
and death, its cause, its cessation and path leading to the cessation of decay and
Te Buddhist Teory of Inference 5 1
deat did s o in the same way as I do at present; all those recluses and brthmm;as
who in the future will thoroughly understand te natue of decay and death will
do so i te sar -e way as I do at present-tis consts te inductive kowledge. 4
In te Abhidhamma and PrkarlQs we find a more favorable attitude toward
the science of debate and logic. Of course, argumentation was always kept
subservient to scriptures. s
As we remarked earlier, te science of logic was developed in India for the
sake of sophistication in debate. Most of te terms and techniques of te science
of debate were retained in te science of logic. I Kathtvatthu we come across
several such terms, lie anuoga, tdhtr;a, patijft, upanaa, niggaha, anuloma,
patikamma, and nigamana.
Yogacartbhamiststr, a later work dealig wit the science of debate,
contains a section pertaining to te doctrine of anumtna and other prmtIas.
Tcci is of the opinion that it is, in fact, there that we find the earliest elaboration
of te doctrine of syllogism and prtlas. 6 Another work by te nanle Tarkatstr
deals more exhaustively wit the teory of anumina and tus forms a link between
te science of debate and science of logic. This Tarkatstra points out unmistakably
the existence of trirapa theory before DiInaga. 7 Tcci observes : "Weter .the
work is Vtdavidhi or a diferent one, tere is no doubt that it contains ideas and
doctrines which DiInaga foud before hinl and which in many places he refted
i his treatises, and in oter places he followed. It is still a vivtda text, but it
shows an enormous progress upon te first attempts and mere catalogues of the
older treatises, as we can see quite well when we compare it with te vivtda
sections contained in Maitreya and AsaIga. " 8
ANNA I T WORS OF PR-DAGA TR
The systematic stdy of logic and epistemology that began i the Nyaya
circle was introduced i te Buddhist tradition by AsaIga and Vasubandhu, who
tried to counter Nagarjuna' s niilistic dialectics. Nagarjuna was an absolutist
who argued for the ftility of all logic for the experience of the Absolute. With
the advent of AsaIga and Vasubandhu the realistic Nyaya logic was introduced in
te Buddhist circle with suitable adaptations to their idealistic famework.
Asarga was perhaps the frst Buddhist writer who talked of the theory of the
partrthinumtna (afctvayavi vika) . However, we don' t fd any innovation or
sigfcant contribution in the works of Asarga. A shor summary of the exposition
of logic by Asarga is found in PrkarIira Vtcisistr and volumes 7 and 16 of
Mahiytntbhidharmasaruktasangiti ststra. Asarga was preceded by
Maitreya, who composed a treatise on the art of debate by the name
SaptadasabhamiSstryogacara. In logical views Asarga followed Maitreya
except in respect of te theory of proof. Proof (sadhaka) was understood by
Aarga in a comprehensive sense of prmaIa and was subdivided as follows :
(1) p
rtjf, (
2) he
tl
,
(
3)
ur1, (
4) lpana
a, (5) n
i,
(6) p
rta,
(7) anlmana, and (8) agama. The frst fve subdivisions constitte anumana.
Asarga differed from Maitreya in diferentiating anumana from prtaka,
52 Buddhist Epistemology
upamina, and agama and in regarding anumana as consisting of the frst fve
divisions only. But in all this he seems to be just iitative of the Nyaya view.
ASaIga was followed by Vasubandhu, who was a renowned teacher of logic.
Vidyabhushana
9
has presented a summary of Tarka sastr authored b Vasubandhu.
He points out tat Vasubandhu seems to have used two fors of anumana, naely,
that of fve parts at the time of debate and a syllogism of two parts, namely,
prtifa and hetu, on an ordiary occasion.
lO
Stcherbastsk, however, mentions
tat mostly he was operating with pafcavayavi vaka, but sometimes he was also
making use of the abridged traavayavi . l He frther poits out that there are
only two other signifcant points on which Vasubandhu differed from the
Naiyayikas : frst, the doctrines of trirupa and avinabhtva appeared frst i the
works of Vasubandhu, and second, the classifcation of hetus and hetvabhtsas is
diferent fom the one accepted in the Nyaya school and agrees in priciple with
the one introduced by Dinaga and developed by Dharmakirti.
AA I T WOR OF DIN AGA
DiInaga gave a new direction and impetus to the study of logic and
epistemology in the Buddhist tradition. It goes to his credit tat he treated the
teory of prmaras separately fom the theor of prmeas. In hi and perhaps
i Vasubandhu prior to hi we frd an atempt to write idependent works on
pramaras witout bringing in explicit analysis of prmeas. According to hi,
there are only two prmazas, namely, prtaka and anumana. The other
prmazas, lie sabda, were reduced to anumtna. Dinaga' s aaysis of anumtna
was so striingly origial tat even in te Nyaya circle it was taken cognizance
of. Uddyotara, for exalple, writes, "Apare t bmvate, nantarlakrthadarsanam
tad vido anumanam
.
" 12
Here the expression " apare" probably refers to Diaga.
The basic point of diference between DiIaga ad the Nyaya logicians is
that for DiIaga anumtna may consist of a pure thought process, tat is, purely
propositional, or it may be sentential . \3 A purely propositional anumana is a
tought process that has not been verbally expressed. It is named by him as
svtrthanumana. Wen the propositions of an anumtna are expressed i language
(rkasana or tkhtna) , ten tey are sentential (sabda) i nature, and tat form
of anumana is named partrthtnwnana. Keit
14
opies tat te distiction between
the to kids of anumtna was frst itroduced by Dinaga i Indian logic. Probably
this two-fold classifcation of anumana ws already prevalent in the Nyaya tradition,
but tere it was used i a slightly diferent sense, svarhtnumtna meanig anumtna
for one' s own sake and partrthanumana meaning for the sake of others. Diaga
borrowed the same phraseology to name his tofold classifcation of anumtna.
So far as pararthanumana is concered, it is sentential, according to Diaga,
and normally we resort to sentential expression only when we communicate with
others. But conmmnication with others is not the necessary condition of usig
language. Thus, Diiaga would not agree wit the Naiyayikas in regarding
pararthanumtna as inference for the sake of others but would insist that this is
one of the motives for udertaking pararthtnunulna. His basic difference with
te Naiyayikas is wit regard to te understanding of svarthtnumana. For the
The Buddhist Theory of Inference 53
Naiyayias swirthtnumtma is also sentential because for tem no conceptual
kowledge can be nonverbal . For Dilaga, on the other hand, there can be a
nonverbalized 'conceptual cognition, and in svtrthinumtna we have only such
cognitions.
The other point o
f
deparre of Dil aga is the advocac of svabhtvtnumtna,
which stands for a sort of analytical entailment . In the Nyaya tradition, because
of its predominant realistic attitude there was no such possibility of the acc
e
ptance
of one concept being subsumed under another concept such that the former could
be deduced fom the latter. In te Buddhist tradition te entire process of anumtna
operates at the conceptal level only, and therefore te Buddhist logicians could
conceive of one concept being conected wit aoter by te relation of analyticity.
Anoter innovation of Dilnaga, which is a corollary of te earlier mentioned
poit, is that he made explicitly clear the distinction between te objects of
perceptual cognition and te objects of inferential cognition. In anumtna we are
concered not with te real object (svalaka1a, grtha) but wit te conceived
object (stmtnalaa1a, adhavasea) . So a tought process fom one conceived
object to another conceived object is necessarily connected with it.
Lastly, in Dilnaga we find a new understandig of the concept of anlnea.
According to te Nyaya logicians stdha is anumea, but for Dilnaga it is paksa
characterized by stdha, which is te anlnea. So in the example of the inference
of fire on te basis of the perception of smoke Dilaga would maintain that the
thought process is from smok hill to fiery hill and not fom smoke to fire. The
latter Naiyayias emphasize the colocushood (ektihikar1ya) of hetu and stdha
i paka, whereas Dilnaga talked of hetu an
d
stdha being the coproperties of
paka (dhamw) .
Anuia in the Work of Dharmakirti
Dilnaga' s formulation of the theory of anumtna was frther explicated by
Dharmakirti in his Prmtna Vartika and Naya Bindu. In the latter work, te
basic concepts and terlS involved in te theory of anlna have been clearly
analyzed and defmed. Dharmaklrti' s main contribution lies in his elaboration of
the doctrine of trinlpya, a detailed account of which is presented later. The
doctrine of trinpa regards hetu or lilga as the starting poit of the process of
anumtma and grounds its validit on three conditions that a valid hetu has to
flfill . In short, they are as follows :
1 . There should b e a necessary presence of hetu in paa (anwnee sattvam) . This
corresponds to the idea of paadhannatva available in the Nyaya position.
2. There should be presence of hetu only i n such cases that are similar t o paka
(sapaea sattvam) .
3 . Lastly, there should b e necessary absence of hetu in such cases. that are dissimilar to
paa (asape ctsattvameva niscitam) .
The second and the third conditions correspond to the two tpes of vtpti ,
54
Buddhi st Epistemology
namely, afnative (anvaya) and negative (vatireki) , accepted in the Nyaya
tradition. Though tese tree conditions of hetu or Zinga have been introduced by
DiInaga, Dhannakirti put fort teir systematic fonulations. In fact, here it is
given such a central position tat the entire subsequent discussion on anumana
pivoted around it.
Further, it goes to te credit of Dhannakirti to have analyzed anupalabdhi
into eleven varieties. Such a classifcation of negative relation is of great signifcance
in the analysis of causal relation. It seems that the later Buddhist theory of
pafcakarli15 was fonnulated as a consequence of the analysis of the negative
aspect of causal relation, as it was put forth in the analysis of the diferent varieties
of anupalabdhi .
After Dhanakirti we don' t come across any notable thiner i Buddhist
logic. Dharlottara, Mokakaragupta, Prajfakaragupta,
S
antarakita, KmalasIla,
Ratnakirti , Jfanasri Mitra, and others explained, elaborated, and defended te
Buddhist doctrines of logic, but none of them made any new contributions. In the
analysis of the teory of apoha Rtakiti introduced some signifcant innovations,
but he didn' t take much interest in te field of knowledge. However, he brought
about greater sophistication and cohesion in the Buddhist system. Thus, it appears
that Buddhist logic of a serious sort began with Vasubandhu, was developed by
DiIaga and Dharmakirti, was defended by
S
antarakita and KmalasIla, and
was later refined by Ratnakirti and Jfanasri Mitra.
Svataf-Parataf Controvers
The problem concering the pramalya of prm a can be posed in two diferent
ways. It can be discussed as to whether anumana is a prma!a or not, or, i
other words, whether anumiti jfana has pramalya or not . On this point, with the
exception of Carvaks, all oter schools of philosophical tought accord pramlya
to anumiti jfana, but i what way anwniti jfana has pramlya tere is a diference
of opinion. Vedantins and Bhartrhari circumscribe te independent operation of
anumana prmalJa by making it subservient to some other prllala. The rest of
te schools and thinkers grant its independent operation. The Buddhists with
whom we are presently concered accept anumana as an independent prmalJa.
However, in DiInaga we meet with anoter position that assigns a status to
anllana that is on a par with illusions, hallucinations, and so on. But such a
stand is to be understood in its proper metaphysical perspective, which talks of
two levels of realit, namely, parmarhika and sarvrtka. The validity of inference
is denied only from the parmarthika point of view, but at te levd of stilvrtika
it is very much valid, as valid as perceptal cognition.
1 6
Wether anUllana i s an independent prmalJa or not i s a question that has a
metodological significance, but a question philosophically more significant is
the mode of knowing prallalya of anlllliti jfana. On this issue Indian
epistemological thinkers are generally divided into two camps. Some tiners
The Buddhist Theory of Inference 55
adhering to svata!prama1yavada regard the relation beteenjiana and, for that
mater, anumitijiana andprama1ya as svata! (intrinsic) . Some others view this
relation to be parata! (extrinsic) . Buddhist thinkers, particularly Dinnaga and
Dharmaklrti, advocate the teory of svata!prtma1ya.
The question aboutprama1ya ofprm is iseparably bound with te question
of the nature of prma. A stated earlier, Buddhists refse to draw a disti<tion
beteen prma1a and prma1aphala, saying tat every cognition-qua-cognition
is ab iitio valid insofar as it is caused b the object itself. The sarupya of the
cognition with its object is a suffcient guarantee for the tth of the cognition,
though tis sarupya is not somethig extraneous to that cognition. For Buddhists
every cognition has twofold aspects (dvairupya) , namely, svakarta and
viayakarta. Viaakarta is the ground for the prama1ya of a prma since
viayakarata is the very natre of a cognition. Every cognition has to be iherently
true. This consideration led these Buddhist thinkers to advocate the theory of
svata!prama1ya.
The intrinsic trth of a perceptual cognition is pria facie itself a very
plausible position, but the itrinsic truth of inferential cognition may not be that
apparent. This is because of the complicated structre of the inferential process.
In every inferential process there is a possibility of going astray at every step,
givig rise to what is technically kown as hetvabhtsa. Further, there is a greater
need for adducing evidence in an inferential cognition compared to the perceptual
one, and this may make the pramt1ya of the inferential cognition contingent
upon the adequac and conclusiveness of the adduced cognition. We may therefore
state and examine the grounds on the basis of which Buddhists insist on the
svata!prama1ya of anumiti jiana. In order to appreciate the sharpness of te
Buddhist position it would be very much helpfl to refer to Vacaspati , 17 the
Naiyayika, who accepts the svata!prama1ya of anumitijiana even i the context
of te Nyaya tradition. His argument of the ab initio apprehension of prama1ya
of anumiti jiana is that among its originatig conditions there is certainty about
the vapti jiana. There is no room lef, therefore, for having any subsequent
doubt in the validity of inference. I should be pointed out that Udayana is not so
very confident, but he, too, hesitatingly concedes that trth is there apprehended
ab initio. However, he also makes a desperate attempt to reconcile this with the
part/ theory of the Nyaya tadition by suggesting that there it may be both.
Gangesa is not oblivious of this inconsistenc with the Nyaya standpoit and
therefore proceeds to explicate what in his view is te true intention of Vacaspati.
The general poit that he makes is that in all these cases there is no scope for
doubt, though truth is apprehended extrinsically. In this he is following Udayana' s
comment on Vacaspati in his Parisuddhi . 1 8 The crux of his argument is that it
would be too much to claim that since there is no scope for doubt, the trth of an
inference is apprehended ab initio. Truth is always to be known subsequently in
anuvyavasaya.
Dinnaga and DharmakIrti seem to be uncompromising in their advocac of
the theory of svat!pramalya. In Prama Samuccaya Dil aga. explicitly maintais
56 Buddhist Epistemology
tat te guarantee for te trut of a cognition lies in its being of the form of its
object . He writes tat, "the means of cognising is sinlply te cognition havig
the fonn of te object. "
1 9
For h every true cognition necessarily refers to an
object, and since it is caused by an object, it possesses te form of tat object.
Dilnaga is a sakarjfanavadin and, terefore, insists on the knowledge b
e
ig of
the form of its object. If every true cognition has to be of the nature of its object,
it is a suficient condition for its truth and terefore leaves no scope for any
doubt. The question of doubt will arise only when tere is a possibilit of te
intermingling of such elements tat are not caused by te object and tat are due
to subjective factors. For Dilaga, therefore, there is no distinction between
knowledge and such factors tat evidence its trut insofar as te latter is not
something extraneous to knowledge. The question of evidence arises only when
falsit is suspected, tat is when someting that is taken to be a prma!a
(knowledge) is suspected to be a prm(1abhasa (pseudo knowledge) . If, on the
basis of any evidence, a particular cognition stands falsifed, it does not mean
that what was knowledge is now falsified. In fact, knowledge by definition is
true, and trut follows from te very defnition of knowledge as a matter of
analytic necessit. Therefore, it is impossible to falsif kowledge. So when on
te basis of some evidence or set of evidences a particuar cognition is falsifed,
all it means is tat a pseudoknowledge tat was masqueradig as a genuine
kowledge is now exposed,
In order to appreciate te strengt of Dirnaga' s position, it is necessary to
first see his metaphysical position. Being an advocate of Vjfanavada, Diraga
maintais tat in the ultimate analysis te entire empirical reality is notig but
diferent manifestations of the. series of vifana, and as a consequence of tis he
maintains that the means of cognition, te
c
ognition tat is its result and the
object of cognition, are not separate from one another, and though in an
epistemological analysis we may analyze these three factors of cognition as
prma!a, prm(1aphala, and prmea, i reality they are not tree separate
things.
20
Following Dilnaga, Dharmaklrti also subscribes to te svatafpram(1ya
teory, tough he talks of a true cognition necessarily leading to a successf
activity.
21
Kowledge leading to successfl activity is only a test and not its
criterion. A distinction, terefore, has to be drawn between test of trut and
criterion of trut. The trut of a cognition can be tested on te basis of extraneous
considerations, lie beig a means to successfl activity. It should not be mistaken
as a criterion of trut. The purpose of a test is to verif or ascertain te presence
or absence of a particular phenomenon. This can be done on te basis of te
possible efect tat a particular phenomenon gives rise to; for example, in
Dharmaklrti ' s system successfl activity is an outcome of true cognition, and,
terefore, the former can be taken as a test of te later. A criterion, on the other
hand, serves a diferent purpose. Its function is to distinguish a particular
phenomenon from its opposite. In te case of knowledge, accordig to Dharmai,
arthasaripya is a criterion of knowledge and is not something extraneous to
The Buddhist Theory of Inference 57
knowledge. It is rather the very nature or essence of knowledge. Thus,
avisGdakam is a criterion of kowledge, whereas sarapurarha siddhi
het is .te test of kowledge.
From tis aalysis it follows tat te criterion of tut of an iferential cognition
is intrinsic to it, and tis provides a suffcient ground for maintaining that Dinaga
and Dharmaklrti subscribe to the theory of svatalpramilJya.
NAT OF ANA
The teml anumina literally means "a knowledge tat follows" (anu+mina,
i . e. , pasctnminamiti anumtnam) . This means that the inferential knowledge is
necessarily preceded by some oter knowledge. In oter words, anumtna as a
prmtw is a complex kowledge consisting of two elements, one antecedent
kowledge and the oter consequent kowledge. The antecedent and te consequent
have a particular tpe of relation tat is technically known as gama-gamaka
bhiva (entailment) . The antecedent leads to, or gives rise to, the consequent and
is, therefore, a gamaka (that which entails) . The consequent results fom the
antecedent and is, therefore, a gama (that which is entailed) .
However, it is not te case that any kowledge will lead to, or entail, any
other knowledge. The antecedent knowledge has to be in the form of lilga
(necessary mark) , of which the consequent knowledge is Zingin (marked) . Linga
means tat which is a necessary mark of someting other than itself (Linamrtham
gamati saI Zingam) . Likewise, lingin stds for tat which is necessarily marked
by a Zinga.
Between lifiga and Zingin there is always a gama-gamaka-bhtva, 22 which
can roughly be regarded as te relation of entailment such that every case of the
presence of Zinga is necessarily a case of te presence of Zingin, and, on the
contrary, every case of the absence of lingin is the case of te absence of Zinga.
The necessary connection betWeen Zinga and lingin on te basis of which gama
gamaka-bhiva is established between the two is technically known in te Buddhist
tradition as avintbhtva or vtpti (necessary concomitance) . The presence of
avinibhiva provides te basis of the transition from gamaka to te gamya.
Avintbhtva, terefore, constittes te logical ground for te process of anumtna.
It is significant to note here tat unike the Nyaya tradition i te Buddhist tradition
Zinga and Zingin are in te form of concepts (vikalpas) rater tan things or
metaphysical reals, a point discussed later.
Nonpresentative Character of Auia
The inferential kowledge, as we have stated, stands for te kowledge of
te lingin on the basis of te kowledge of the Zinga. This iplies tat te ierential
kowledge of te lingin is necessarily nonpresentative. By nonpresentative is
meant that the object of cognition, which is the lingin, is not directly given to, or
apprehended by, the cognitive senses. The object here is not directly present but
is conceived to be so on te ground of the presence of its necessary mark. However,
it should not be understood that te conceived object is essentially a nonexistent
58 Buddhist Epistemology
one. It is an existing object whose presence or absence is indirectly cognized at
a particular place on the basis of the presence or absence of its mark. Thus,
anumana as a prma1a offers a mediated knowledge of an object that is not
directly given. We know an indirectly given object trough a direct one to which
it is related.
In saying that anumana is nonpresentative it is to be contrasted with
prtaka, which is essentially presentative. Presentativeness of perceptual
cognition consists in its apprehension of an object that is directly given to the
senses (aam akaamitiprtaam) . Such an object is named as graha (sensed)
i contrast to te object of inferential cognition named as adhavasea (conceived) .
The graha, svaZaka1a, is an objective reality, which is just a point instant, a
unique particular, according to Buddhist ontology. The adhavasea, samana
Zaka1a, on te contrary, is a mental construction (kaZpana) , a conceived entity,
conceived in terms of eiter its substantiality or objectivit or relation or generic
character. This conceived object is neither purely objective nor purely subjective
but intersubjective. It is not objectively real but hypostatized to be so. However,
its hypostatized character does not make it illusory or purely imaginary. It is a
construction out of sense data and thus has empirical realit.
Judgental Character of Anuia
According to the Buddhist epistemology as stated earlier, tere are only two
tpes of knowledge because there are only two tpes of objects of knowledge
(manam dvividham mea dvaividhtt) . Prtaka is kowledge in the form of
pure sensation devoid of all conceptualization and judgmentalization
(rtakam kaZpanapoham) . In contrast, anumana is necessarily a kalpana;
that is, conceptual, ad hence judgmental in nature. Thus, in the Buddhist tradition
a knowledge can be either non judgmental or judgmental . If te former, it is a
case of prtaka, and if te latter, it is a case of anumana. Of course, every
judgmental cognition is ultimately grounded in a non judgmental cognition, and
that is why anumana is characterized as prtaka prtha bhavi .
In the Buddhist tradition anumana is not the onl tpe of judgmental cognition.
An illusory cognition (bhranta jfana) can also be judgmental , but it is not a
prma1a, tat is, true cognition. So the only judgmental cognition that is a
prma1a is in the form of anumana.
Relational Character of Anuia
Unlike the perceptal cognition, inferential cognition is relational in natre.
The object ofthe inferential cognition is always a characterized entity (dhamu)
that is cognized on the basis of the perceptal cognition of one of its dharmas. A
we pointed out earlier, there are two distinct types of cognition involved in an
anumana, namely, the antecedent and the consequent. Both these cognitions are
relational in natre. The antecedent cognition is again tofold, technically kown
as pakadharmatva and vapti or avinabhava. The forer is in the form of
The Buddhist Theory of Inference 59
contingent relation, but te latter is a necessary relation. There is an involvement
of relation not only i the structre of the premises and the conclusion, so to say,
of the anumtna, but also in te transition from premises to conclusion. This
relatio
n
makes possible an extension of our knowledge from premises to
conclusion. Thus, anumtna operates within a relational framework through and
through.
According to Buddhist metaphysics, all relations are mental constrctions,
as there are no relations in realit. So relations are not apprehended ilprtaka.
That is why perceptual cognition is nonelational in nature. Thus, only in anumtna
are relations apprehended. The objects of anumana are relational , and anumana
in itself is also a relational cognition.
Verbalizable Character of Anuia
There has always been a difference of opiion regarding te relation between
kowledge and language. Bhartrhari , for example, regards every kowledge to
be necessarily embodied in language. Buddhist thiners, on the other hand, insist
that tere can be thought even without being associated with language. They talk
of two tpes of kowledge; tat which can be verbalized and that which canot
be verbalized . Perceptual knowledge is nonverbalizabl e. In fact , its
nonverbalizability follows from its being nonjudgmental or nonconceptual .
Inferential kowledge, on the oter hand, is verbaliable. It is contoversial whether
it is always verbalized or whether it is so only in partrhanumtna. However, one
thig is certain, that whether actually verbalized or not, it alone is verbalizable.
In fact, only when the inferential cognition is verbalized can its fallacious or
nonfallacious character be exposed.
Generalized Character of the Object of Anuia
The perceptual cognition has the unique particular as its object. But the
object of inferential cognition is a generalized concept that results fom the process
of mental construction. The constructing activity of intellect is dichotomous. It
always begins by conceivig the object in two aspects, the sinilar and te dissililar.
It opera
t
es with the method of agreement wit te sinlilar (anvaa) and difference
with te dissililar (vatireka) . If the aspect of agreement is expressed, then the
aspect of disagreement is understood, and viceversa. Thus, according to
Dhannakrti, te object of inference is the exclusion of te opposite (atvtvrti) .
Anumana does not cognize the positive nature of a thig but only its negative
natre, which consists ill a ting' s exclusion fom all types of dissimilar things.
What is meant by saying this is only that the object of anumana is a mental
concept, a universal and not a real existence. However, this universal is not
merely a creation of mental inlpressions (vtsant) . It is generated by real things.
Accordig to Dharmaklrti, it is in the nature of real things to generate a mental
concept. So, a mental concept is invariably related to those tings. Because of
this, though an inference cognizes a mental concept, yet it enables us to attai a
real thing. Insofar as it does not help attaining a real tig, it is bhrantajitna.
60 Buddhist Epistemology
But so far as it enables us to atain an efcient real thng, it is aprnl. Dhaa
again proves that an inferential judgment is ultimately grounded in a real thing.
The real fre that is te cause of real smoke is the remote object of the inferential
judgment "There is fire on the hill . " The real fre generates the real smoke, the
real smoke generates its own sensation, and tis sensation generates the judgment
based on conception that there is smoke on the hill . This judgment with the help
of memory of te concomitance beteen smoke-in-general and fre-in-general
generates the inferential judgment that "tere is fre on the hill . " The actual
object of inferential judgment, is the fire-in-general , which is necessarily related
to real, unique fires.
RELATION BETWEEN PRT$ A AA I TH
CONTEXT OF TH TORY OF PRA- VlST
DiInaga in te begin g of the Prma1a-Samuccaa draws a radical and
essential distinction between pure sensation and a conception. The later, in his
system, includes judgment and iference. This distinction is grounded in a more
basic distinction between two types of objects of cognition, namely, svalaka{za,
the unique particular, and samanalaa{za te universal. The former is a paricular
individual that can never be generalized or conceptualized, and the later is te
universal tat is conceptually constructed by te mid through generalizing from
many individuals without any regard to te peculiarity or individuality. The former
is real , While the later lacks reality. A one is incompatible wit the oter, there
cannot be anything that possesses bot svalaka1a and samanalaka{za at te
same tie. Corresponding to tis essential distinction between the two kinds of
premeas is a radical distinction between the two prma1as. Dinaga was
ucompromising in arguig for te distinct spheres of operation of tese two
prma{zas, and his teory is kown as prma1a vavasthavada.
It is in contrast wit the theory of prma1a sampalava advocated by the
thinkers of the Nyaya and Mimamsa schools. According to them, the same object
can be cognized in many ways by diferent prma1as. There are no strict, distinct
limits for each of the prmi{zas. The basic contention underlyig te teory of
prmiza vavasthi is denied by the Naiyayis, who admit neiter tat tere are
only two quite distinct types of prma1a nor that tere are ony two quite distinct
types of prmea. According to tem, prma1as are four il number, namely,
prtaka, anlnana, sabda, and upamina, and the objects can also be of three
kinds, te paricular, te universal, and the individual thing as possessor of
universal .
The mai Buddhist argument against the theory of prma1a samplava is that
if prtaka and anllmana are admited to have one and the same thing for teir
object, the cognition produced should also be of the same nature; but that is
simply absurd. The sensations of burning as apprehended in prtaka and as
kown through words in anllmana are quite different. The Buddhist thiners
maintain tat the cooperation or mixtre (samplava ekasmin viae saream
prma1inam prvrt / of the different sources of our kowledge in the cognition
The Buddhist Theory of Inference 61
of one and the same object i s impossible, since each one has its own special field
of operation.
Thus, Diimaga and his followers draw a dear-cut difference between a
perceptual knowledge and a conceptual knowledge. Even the perceptual judgment
of the form "This is blue" is treated as conceptual and therefore inferential . In
the strict sense, therefore, all perceptual judgments are actUally inferential .
Diiaga insists that an object of knowledge must be either a svalakafa or
stlnanalakafa, for there is no third kind. If it is a svalakafa, it would be an
object of perception, ad if it is stnala(za, it would be a object of afl.
Just a praa calot coglize a samanalaafa, so alsQ anwnana cmmot cognize
a svalakafa.
Prtaka is capable of cognizing an object only because it is produced
through the efcienc (samartha) of an object. Such a ti alone can be an
object of prtaka, which is able to attribute its defnite form (niyata pratibMsa)
to te cognition. It can be done only by unique particular, which alone is cognized
i prtaka. On the other hand, te unique particular can never be cognized in
anlmana, which results fom the ascertailmlent of invariable relation beteen
two entities. Such a relation Ca1 ot be established between two unique particulars,
and therefore the latter cmmot be the object of anlmana. Moreover, prtaka,
which is in te form of StilllUlus comillg from an object, can be produced only by
such an object that is present. A universal cannot produce such a result because
it can neither cause a stiIulus nor call forth its prtibMsa, since it is altogether
devoid of any kind of direct causal efficienc. According to Buddhists, prtaka
can apprehend only such an object tat is objectively real . This alone is pure
reality il the ultimate sense, the thing-in-itself, because the essence of reality,
according to Buddhists, is just the property of being causally effcient.
That a unique particular of tis kind should also be cognized by the conceiving
faculty of our mind or by anlmana is ilnpossible. The sphere of unique particulars
is not the sphere of anumana. The latter cognizes relations (grhita prtibanda
hetukam) . In a unique particular no relations can be found. Relations are always
conceived between two (or more) universals. The universals are not realities ;
they are logical constructions produced by our constructing mind, and, therefore,
their objective existence in te extemal world belongs only to te domain of our
illferential knowledge, which is either a dialectical superstructure upon reality or
ml objectivized image. Their source, indeed, is not positive, since they always
contain a corelative negation (ana-vavrti OF anapoha) .
A universal is, therefore, illtemal (abtham) mental construction but related
to extemal reality, since ill our behavior we don' t notice te differences; that is,
we don' t think tat "a cow" (conceived cow) is not an external object. It is
deemed to be external (bahatvena avaslyamanam) and thus gives reality to
illferential knowledge. Being tus indirectly related to extemal reality, it has
efficienc. Thereby it becomes a consistent experience and hence a prmafa.
There is no oter prmafa except these two. Watsoever has a claim to be a
prmafa is included in tese two, or, if it is not included in them, it is not a
62 Buddhist Epistemology
prmilQ. Since there is neither any other source of valid kowledge nor any
other object to be cognized, one has got to admit the theory of prmila vavasth.
ANALYSIS OF T NATUE OF IERENTIL JGMNT
Buddhist tiners sometimes draw a distinction between a perceptual judgment
and an inferential judgment. All cognitions naturally lead to judgments, tat is,
to interpretation of sensations in concepts. Since a cognition can be distinguished
as direct and indirect, the judgment can also be divided into a direct one and a
indirect one. The direct one is a synthesis between a sensation and a conception,
and te idirect one is a snthesis between a sensation and two concepts. The
direct one has two terms; the idirect one has tree terms. The direct one is of the
form, "Tis is blue" or "This is smoke, " and the indirect one is of te form,
"There is fire ' because there is smoke. "
There is a basic difference between a perceptual judgment and an inferential
one. The perceptual judgment is a direct cognition in te sense that its object is
directly known. "This is smoke" is a perceptual judgment because smoke is
directly perceived. The inferential judgment is indirect insofar as its object is not
directly perceived. It is a cognition of a nonperceived object tough a perceived
objectthat is its mark. The unperceived object has a mark, and this mark, in its
turn, is te mark, of a svalakalQ. Thus, fre is the unperceived object whose
mark is smoke, and tis smoke is, in tur, a mark of some svalakm:ta, te real
smoke. The cognition of a svalakaQ as possessing the mark of its mark is
anumina. In a perceptual judgment we cognize the object "X" through its mark,
which is the concept "B. " In an inferential judgment we cognize te object trough
te mark of its mark, tat is, through "A, " which is a mark of "B. " "A" and "B"
are related as reason and consequent. When te reason is cognized, the cognition
of the consequent necessarily follows. No doubt "A" and "B" have a conmlon
substratunl "y" (aa) ; its presence will be necessarily understood witout any
formal expression. In tat case te two interrelated elements "A" ad "B" will
represent te whole inferential judgment. This judgment will ten apparently
consist of two concepts only but related as reason and consequent, one being te
necessary ground for inferring te oter.
Strictly speaking, there cannot be anything like perceptual judgment because
the theory of twofold prmi1as is based on a dichotomous distinction between
the perceptual , which is non judgmental , and the conceptual, which is judgmental .
In perceptual cognition there is no involvement of judgment, and hence tere
cannot be anything like perceptual judgment . However, i a loose sense we can
talk of perceptual judgment insofar as every perceptual cognition is capable of
leading to a judgment. But ten two questions arise, nanlely:
1 . Is this perceptual judgment the same as, or diferent from, anlllntna?
2. I f i t is diferent fom anwntna, can i t be a prmt(w?
Arcata, 23 conmlenting upon Dharmaklrti , writes that Buddhist logicians, in
The Buddhist Theory of Inference 63
fact, don' t consider perceptual judgment to be a case of anllmtna and don' t
regard i t to be a prmtla. The perceptal judgment i s not a prmtla because it
is a cognition' of what is already cognized .(gfahiagrthz) . Every perceptual
judgment immediately follows in the wake of perception, and, therefore, it
apprehends that very object that is apprehended in perception. Anumtna, on the
oter hand, cognizes an object hitherto uncognized (agrhlagrthi) .
There is another reason tat perceptual judgment as distinct fom inferential
judgment is not a prmtla. Though the perceptal judgment has the same object
that is already sensed (perceived) in prtaka, while determining it, perceptual
judgment distorts it, that is, finds it as something general though, in fact, it is
something unique. Every judgment consists in superimposition of character(s)
on a ting that it doesn' t , i fact, possess. The inferential cognition, on the other
hand, is conceived only wit that object that is a concept. Since every concept is
a mental construction, there is no question of its being distorted by mind.
It then becomes a judgment of concomitance or a judgment based on
concomitance. Inference or the object cognized in te inference, says DhamlOttara,
is a complex idea of te substratm together with its inferred property, or, when
the invariable concomitance between the reason and the inferred property is
considered abstractly, then the .inferred fact appears as this praperty taken in its
concomitance wit te reasan. 24 In the first case, we have just an inferential
judgnlent, and in the secand case a judgment of cancamitance. The fann .f the
farmer will be, "There is fire there because there is smake, " and tat .f the
latter, "Wherever there is smake, there is fire. " The term "there" is expressive of
te substratm (aka) where the two attributes, namely, smake and fire, are
necessarily interrelated such tat the latter is deducible from the former. Thus, in
every inferential cagnitian there are three temlS invlved. Hawever, nat every
cagnitian cantaining tree terms .f which .ne is te substratum .f the ater twa
will be an inference. Only such a cambinatian of tem, where twa attributes are
necessarily interrelated, the .ne deducible fom te ather, represents an inference.
DEFIION OF AA
Before attempting ta give a definitian .f anumtna, DiJnaga, Dhannaklrti ,
and their fallowers begi by drawing a distinctian between twa kinds .f anumtna,
namely, svtrtha, which is a mere mental process, and partrtha, which is a
verbalized farm .f that mental pracess. Since the frst .ne is nanverbalized, and
the secand .ne is necessarily verbalized, the t cannat be cainprehended under
.ne single defnitian, and, terefare, before attempting ta a give defnitian .f
anllmtna, it has ta be specified whether it is a defnitian .f svtrthtnllmtna .r
that .f partrnumtna. This paint is clearly stated by Dharmattara in his
cammentary an the Ntya Bindu. He writes, 25 "Partrthtnumtnam sabdttakam
svtrthtnumtnam tu jftntakam
.
Tayortantabhedtnnaikam lakawmasti.
Tatastayo(l prtiniyatam laka{zamtkhytm prkarlabheda(l kathate. "
Thaugh twa separate definitians far each type are given, we find that the
64 Buddhist Epistemology
differential propert of anumana remai the same in both cases. Diferent types
of definition of anumana are available in the Buddhist tradition. The earliest
attempt to define anumana seems to have been made by Vasubandhu, who in the
Vada Vdhi defnes it as "Nantarfakrtha darsanam tdvido anlmanam. "A siilar
defnition is put forth by DiInaga. This defnition lays stress upon the inseparable
connection tat unites the hett with the sadhya and defines anumana as a
knowledge arrived at on the basis of that inseparable cOlmection by a person who
is acquainted with that cOlmection. The term nantarlyaka or anantarlyaka means
necessary nonexistence of one object in the absence of anoter object. The
Buddhists use another expression, namely, avinabhava, which also means the
same. The doctrine of trirpalifga is nothig but an explication of the concept
of nantarlyaka. It implies the presence of hetlt in te paka, its presence in the
sapaka, and absence in the asapaka because in the absence of such a situation
there cannot be inseparable connection between hem and sadha leading to
inferential knowledge.
Keepig the concept of trinlpyaliiga as the focal point, Dinaga26 puts
forth another defnition that has been given prominence by Dharmakrti and that
finds greater acceptance in the tradition. In the Naa Bindu27 Dharmakirti
expresses it as " Trirpallingtd yad anumee jianam tad svarthtnumanam, " that
is, tis defmition is based on the actual process tat originates iferential cognition.
It is a fnctional definition on a par with a similar fnctional definition put fort
by te Naiyayikas, namely, " VaptiviS!a pakadhannatajianajanamjianam
anumiti . " Concering the Buddhist definition of anumana, which is on te basis
of trinlpyalifga, Dharnlottara28 remarks that it is put forth keeping in mind the
origin of anllmana. The inferential cognition arises out of the cognition of the
threefold mark. Thus, the cognition of fire (which is not perceived) arises out of
the cognition of smoke.
Stcherbatsky29 refers to another definition given by Buddhists based on the
nature of object apprehended in anumana. Anllmana is the cognition of an object
that is not revealed to the senses. All objects can be divided into present and
absent. The present are cognized by perception and the absent by inference.
In Dharmottara30 we also meet with the etymological definition of
anumana. He defines anumana as "lifgagrhmJa sambandhasmarwzasya
pascannnamanumanam. Here anumana is taken to be that cognition that is
implied by the perception of the lifga that characterizes te paka and the
remembering of the necessary concomitance between the hetu and the sadha.
Defnition of Pararthiuia
The onl y fundamental difference between svarthanumana and
pararthanumana is that the latter consists i n linguistic expression of te former. 31
Since the expression aspect is the basic point of distinction in the definition of
pararthtnumana, this aspect has been highlighted. Thus Diiaga defnes it as
"Pararthanumanam tt svadr.tartha prkasakam,, 32 that is, pararthanumana sets
Te Buddhist Teory of Inference 65
forth an object tat has been apprehended by oneself. DharmakIrti also defnes it
as " trirpalinga akhyanam pararthanumana.
"33
Akhanam is the same as
"prktnam, " and every akhana or prkasana is through vacana. That is why
Dharmottara writes, :' Vacanena hi trirpam lifgamakhaate.
"34
Keepig this i
mind, he describes paranhatlllana as sabdatmaka and svarthanlmana as
jfanafaka. 35
To conclude, in te Buddhist tradition we have a variety of defmitions of
anumana put forth on the basis of different considerations. Here we fid some
defmitions based on etymology and some based on te actual process of anumana.
These definitions are substantially different from te ones given in the other
schools. Whatever differences are tere are due to their divergent ontological
positions. All agree tat in every inference tere are two basic elements, namely,
pakadhannatva and vapti, which must feature in any fnctional definition of
anllnana. Similarly, if any defnition is put forth that keep te object in view,
ten it has to take into accout te paroa (concealed) nature of the object
whose necessary mark alone is a perceived object . Likewise, i any defmition of
te etymological type te inplicative nature of te perceptual cognition has to be
brougt to te fore. However, te basic feature of te Buddist defmition from
the fctional point of view is the concept of trinlpalinga, which is insisted
upon as a necessary prerequisite of every valid anllmana.
CONSTIN OF
A
A
According to the Buddhist system of logic, like other systems of Indian logic,
the process of anumana involves three basic terms and their interrelations. This
is quite evident from the analysis of the following defnition of anlmana given by
Dharmakrti :
TrirupiliJgid yadanumee jfinam tad svirthi nwninam
36
These three terms are paka (logical subject) , sadha (logical predicate) ,
and hetu (reason) or linga (mark) , which logically connects paka and stdha.
A analysis of tese constituents of anumana and teir interrelations follows.
Pa
Paka stands for te subject under consideration i the inferential reasoning.
Every inferential reasoning pertains to some individual or class of individuals37
about which we want to infer or establish someting. It is teclically known as
paka. Etymologically, paka (acate iti paalJ means tat to which hem and
sadha belong as its properties . In tis sense it is also referred to as dhannin, the
underlying substratunl, to which hem and sadha are ascribed as dhannas. It is
also nanled as anumea because it is an object of inferential inquiry. It is not a
mere reference to te bare paka but to that paka that is a dhannin and to which
het and sadha are the dhannas. That is why Dharmaklrti defnes paka or
66 Buddhi st Epistemology
anumea as jifasita viSeo dhannf. 38 Though at the level of objective reality
(armanha sat) Buddists don' t entertain the distinction between dhanna and
dhannf, at te conceptual level , which underlies all ou worldly behavior, such a
distinction is very much necessary because no conceptualization can take place
witout bringing in distinctions in terms of drva, gUa, kanna, and jtti .
Wth regard to te ontological status of paa Buddists fndanlentally differ
from Naiyayias. For Naiyayikas, paka and so on, stand for te objects that are
objectively real . They don' t grant any distinction between objectively real and
conceptually real . But in the Buddhist system, there being a clear-cut distinction
between the objectively real and the conceptually real , anumana is said to pertain
to te latter only. Thus, for exanlple, in the anumana "This mountain possesses
fre because it possesses smoke, " the expression "this mountain" is paa insofar
as it possesses smoke and fre as its dhamlGs. Wen we use the expression
"this, " it should not be Uderstood as referring to some svalaka{lG. It refers
only to the conceptualized experience of some svalaka1a. Since paka may
stand for an individual or a class of individuals, a distiction can be drawn
between two types of concept corresponding to a paa, namely, a concept having
universal denotation and a concept having individual or particular denotation. If
te concept of "al men" is te paka in an anumana like "All hUlan beings are
mortal , and therefore they are living beigs, " ten te term "all men" is the
paka, which has a universal denotation (sakaladesavrtitva) . But in an anlmana
lie, "This hill possesses fire because it possesses smoke" te expression "te
hill " is paka, whos e denotation is confined to an individual obj ect
(ekadesavrttitva) . Thus, a paka may have a generalized denotation or a
particularized one.
From anoter point of view, we can draw a distinction between "time-bound"
and "tinle-free" paa. A paka in svabhtvanlmana is tinle-free in te sense
that the dlzannita (property possessiveness) of paka in respect of hem and sadha
is not restricted to any particular time. For eample, when we say, "It is a fower
because it is a rose, " the properties of "roseness " and "fowerness " as they are
ascribed to te object stand in a relation that is a matter of analytical entailment.
The concept of " fower" is deduced from the concept of " rose" as a matter of
analytical entailment, irrespective of any consideration of time. But te relationship
of taltatti (causalit) stands on a diferent footig. It is not a matter of analyticit
to arrive at the concept of "fre" from te concept of " smoke. " Watsoever
analysis we may give of the concept of smoke, it will not entail te concept of
f. I is because te establishment of tis relationship is not a matter of linguistic
stipulation but someting tat is grounded in our experience. It is not that the hil
always has smoke ad fre as its (dlzamw) properties so tat only at tose moments
when te hill possesses smoke, can it be inferred tat te hill possesses fre.
Thus, the relationship of smoke and fire vis-avis hill is time-bound, whereas the
relationship of te object that is te referent of te word " it" vis-a-vis " rose and
fower" is tinle-free.
The Buddhist Theory of Inference 67
The epistemic status of pak$a in the Buddhist analysis is quite different from
tat of te Nyaya analysis. In the early Nyaya tradition pak$a is understood as
sandigdha sadhawln, tat is, tat in which te presence of sadhya is suspected.
Thus, pak$a has te stats of being doubtfl. In te latter Nyaya tradition pak$a
may have any of te following tree statses :
1 . Having an assured existence of sadlra and yet there being an inquisitiveness to prove
this.
2. Having no assuredness about the existence of sadhya and there being no
inquisitiveness to prove this.
3 . Having no assuredness but there being some inquisitiveness to prove it.
The diference between te early Nyaya position and te Navya-Nyaya position
is that whereas in the former the inferential inquiry begins without doubt, i te
latter three different possibilities are accepted in which assuredness and
inquisitiveness in their diferent combinations in ters of teir positive and negative
relations are viewed.
In Buddhist logic, however, instead of doubt (sandeha) being the starting
poit of inferential inquiry, inquisitiveness (ifta) is put forth as te chief motive.
This inquisitiveness leads to a desire to prove. It seems that the idea of
inquisitiveness of Buddhists enabled te Navya-Nyaya logicians to put forth the
idea of will to prove (si$adhai$t) . In the course of development in the theory of
anllmtna we find a logical transition fom sandeha to jiftst and fromjifasa to
si$adhai$t.
Role of Pa$a
Pak$a performs a role that is equally signifcant with the role performed by
hetll or linga. It is the perceptual cognition of te object whose conceptualization
is pak$a, which provides the starting point of te inferential inquiry. Possession
of hetl by pak$a, technically known as pak$adhamzata, is thus the necessary
condition for an anlmtna to take place. The significance of te role of pak$a can
best be understood when we talk of it as a dharl. If there were no pak$a at the
back of which there is a perceptal cognition of some object, the entire process
of inference would have been reduced to an abstract intellectal exercise having
no relevance to te empirical realities. Therefore, te eXistence of pak$a having
some objective reality at its back provides te ground or basis for the inference of
stha. If the pak$a were such a concept that is barren or empt, having no
reality to fall back upon, it would not provide an tsraa (asis) to hem and
stdha, and it would ten be a case of fallacious inference. Not only should the
pak$a not be a barren or empty concept, but it should also be such with. which
the concepts of hem and stha should not be incompatible. Thus, for example,
lake canot be a pak$a in that case where smoke is te hem, and fre is stdha.
Though te Buddhists don' t insist on describing the paka as being the common
substratm (ekthikarlya or samntdhikarlya) of hem and stdha, tis idea
68 Buddhi st Epistemology
of being te substratm and havingpak and sidha as its dhamzas is not absent
in the Buddhist tradition.
Hetu
The other term involved in the process of anumana is hetu or Zinga.
Etymologically, hem means that which takes to an object that is not directly
given (Hinotti gamaat paroarhamiti hetJ) . Similarly, Zinga means tat which
leads to the object not directly given (Park$trtho lingaate gamate 'neneti
lingam) . It is also kown as sidhana because it is a means for te kowledge of
sidha " (sidhate aneneti sadhanam) .
Hem is te pivotl element i the process of anumana. It is a necessary mark
that leads to the inference of the marked object. In order to do so, a hem has to
satisf three formal characteristics ; only then it is kown as sadhem, and it acts
as a suffcient reason for the inference of its marked object.
Hem is simultaneously related to both pak$a and sadha. It is a dhara of
the pak$a and a Zinga (mark) or vapa (pervaded object) of the sidha. In this
respect it corresponds to te middle term of Aristotelian logic, the role of which
is to connect the minor and the major terms. Here also the dhannata of sadha in
respect of pak$a is established on te basis of hem's, being a dhamza ofpak$a.
Diaga, as quoted in te Naa Vartika Tka, 39 defmes hem as tat property
of pak$a that is pervaded by the stdha, which also is a property of the pak$a
(sadhadhamw samanena vapto hem/) . In tis defmition three tgs are pointed
out:
1. Hetu is a paladharma; that i s, hetu is a propert of pala and must be present i n
pala.
2. H
e
w i s the grahyadharma; that is, the dharmata of hetu i s directly apprehended.
3 . Hew i s one of the dhannas of the pala (alasa Gna dhanna) because the
pala also has sadha as another dharma.
On te basis of these tree qualities, the three rpas of het have been
formulated, which are discussed subsequently.
Tpes of Hetu
According to Buddhist logicians, there are tree types of hem, namely,
svabhtva (identity) , ktra (efect) , and anupalabdhi (noncognition) . Svabhtva
and kara are afrmative, whereas anupalabdhi is a negative kind of hem. Now,
let us tm to teir analysis one by one.
Svabhiva Hetu (Reason as Identity)
Svabhtva hem is defined as te one whose mere existence is sufficient for
te establishment of sada. 40 For example, i te judgment "It is a fower because
it is a rose, " the hem, namely, rose, is sufcient for proving the sadha, namely,
The Buddhist Theory of Inference 69
fower. Here the terms "rose" and "fower" have one and the same object for
their reference, though they may have diferent meanings. This sameness of
reference is kown as tadatmya (identit) and is responsible for the existential tie
(svabhva pratibandha) beteen rose and fower. The ter "fower" contains
the extension of the term "rose, " and the term "rose" is a subalter (vapya) , of
which the term "flower" is a superalter (vapaka) . Both are ten said to be
existentially identical and become subject and predicate of an analytical judgment.
Kara Hetn (Reason as Efect)
The second type of hetu is ktra hem or ktra linga, which is in the form of
an efect. It necessarily presupposes its cause. Thus, ktra as linga necessarily
leads to its ktrla, which is the lingin. 41 The relation between seed and sprout,
fire and smoke, and so on is a relation of causality. This relation is given to us i
our experience. We always find tat whenever there is smoke, tere is fire, and
whenever tere is no fire, there is no smoke. The causal relation is a relation of
succession in contrast to te relation of identity, which consists in simltaneity or
coexistence.
The apprehension of necessary connection between a cause and an efect is
based on perceptual experience, but te relation as such is a matter of mental
construction. In te Nyaya tradition it is accepted tat te process of anumtna
may proceed from cause to efect and effect to cause, both ways. That is why
they draw a distinction between purvavat anumana and seavat anumtna, the
former based on te ktra hem and te latter on te ktrWla hetl . The Buddhists,
however, regard the ktra hem alone as sad-hem ad not the kar{la hetl because
wherever tere is an efect, tere must be a cause, but it is not necessary tat
wherever there is a cause, tere must be an effect .
AnnpaJabdhi Hetn (Reason as Noncognition)
The svabhiva hetll and ktra hetu are positive, as stated earlier, insofar as
they help in proving the existence of a thing. The third kind of hetu , known as
anllpalabdhi hetu, is negative in natre. It helps in provig the nonexistence of
a thing. Anupalabdhi has been defined as the noncognition of such an object that
otherwise flfills the conditions of cognizability; for example, a jar is an object
that flfills the conditions of cognizability. If in a particular place there is
noncognition of a j ar, this enables us to infer its nonexistence. So, here
noncognition of the jar is the linga, and the nonexistence of the jar is the lingin.
The noncognition (of a thig) is to be regarded as te linga for the nonexistence
(of that thing) as a lilgin, on the ground that if the thing were present, it would
have necessarily been perceived when all other conditions of perceptibility are
flfilled. Since in spite of the conditions of the perceptibilit being present, if the
thing is not perceived, we can legitimately infer its non existence. 42 For example:
Thesis : On some particular place there is no jar.
Reason: Because it is not perceived, although the conditions of perception
are flfilled.
70 Buddhist Epistemology
A "particular place" is a place before te eyes of the cognizer. But not
every . place is a "particular place" that is present before the cognizer. Now, it
can be questioned, How is it possible for a jar to be perceptible in a place where
it is absent? It is said to be perceptible, although it is absent, because its
perceptibility is imagined. We imagine tis object in the following way: " If it
were present on this spot, it certainly would have been perceived. " In tis case
the object, although absent, is e-hypothesi (samarpya) visible. It is the object
whose absence is cognized, even though all the causes for its perception are
present. We can judge that te causes for te perceptibility of the object are all
present because we perceive other objects in the same act of cognition but not
that particular object. Indeed, when two such objects are before us, we cannot
confine our perception to one of them since there is no difference between them
as regards possibility of perception (rtaka yogata) . Therefore, if we actually
perceive only one of tem, we naturally imagine that if te other were present,
we should likewise perceive it because the totality of necessary condition is
flfilled. Thus, sometig fancied as perceptible is imputed. The noncognition
of such an object is called negation of hypothetical perceptibility.
Bas i cal ly, there is only one kind of anupalabdhi , known as
svabhavanupalabdhi, but in the PrmalJa Vanika43 it has been classifed uder
the following four kinds : ( 1 ) Virllddhopalabdhi, (2) Virlddhakaropalabdhi ,
(3) KarlJanliplabdhi, and (4) Svabhavanupalabdhi. In the Naa Bindu seven
more kinds are added to them, tus making tem eleven in nuber. These seven
are as follows : ( 1) Svabhavavirddhopalabdhi, (2) Karanupalabdhi, (3) Virud
dhavaptopalabdhi, (4) Vapakaviruddhopalabdhi, (5) KarlJaviruddhopalabdhi,
(6) Karaviruddhopalabdhi, and (7) Karalaviruddhakaropalabdhi . 44
Mokakaragupta has added five more to these, making tem sixteen in number.
They are as follows : ( 1 ) Vapakavirddhakropalabdhi, (2) Svabhavaviruddha
vyaptopalabdhi , (3) Karaviruddhavyaptopalabdhi , (4) Karlaviruddha
vaptopalabdhi, and (5) Vapakavirddhavaptopalabdhi . 45
Sadya
The third element involved in te inferential process is stdha or lingin. It
is the property (dharma) that is to be inferred in a paka. It is that property of
the paka that is cognized through anumana on the basis of the perceptual
cognition of te oter propert, earlier referred to as linga. Though sadha may
be expressed as a substratum, for example, fire, and it is so in the Nyaya analysis,
in the Buddist analysis with respect to paa, it is a property, say, the "feriness "
of a given place. In anumana it i s the stdha, the propert belonging to paka,
which is inferred. The sadha does not have a status independent of paka. So in
the Buddhist . tradition it is not the sadha that is the anumea, nor is te
paa b itelf aruea, but te anumea i tat st tat i a dnna of paa.
Aumeya (Object of Iferentia Iquiry)
In te Nyaya tradition sadha is given an idependent status, and tat accounts
for its being the object of inferential inquiry. So in the logical process, it is more
The Buddhist Theory of Inference 71
significant tan pak$a. Only in epistemic terms it has a status lower than the
status of pak$a isofar as its exisntece is either doubted or is intended to be
established. But in the Buddist tradition te iniportance of sadha is, so to say,
passed on to pak$a. Even then, te role of sadha cannot be minimized because
it is te cognition of sadha being te property of pak$a, which is the concluding
part of inferential inquiry. So if it is asked what is cognized in anumana, the
answer will not be that the linga is cognized because wit its cognition te
inferential process starts. Nor can it be said that pak$a is te object of cognition
because the pak$a is to be cognized either wit the linga or wit te lingin and
never witout tem, and its cognition with linga is the starting point. So, te real
object of inferential inquiry is te sadha as it is possessed by the pak$a, which
is technically known as anumea.
A proposition that is unproved and is desired to be proved makes inference
possible. But what exactly is to be inferred or proved? To make the point clear,
let us suppose tat te proposition "The hill has fre on it" is to be proved; but
te hill is already known trough perception, and hence te question naturally
arises as to what exactly is here sought to be inferred. It is not quite impossible
to fd persons who would declare te hill as such to be te anllmea. But conmlon
people naturally think that, tle hill being already perceived, what remains to be
inferred or proved is just the fre. This is te second alternative. We get two other
alteratives as we put major emphasis on either the subject (S , hill) or predicate
(P, fire) , and again we get a tird one if we put equal emphasis on bot subject
and predicate, tat is, hill qualifed by fire or, in oter words, fery hill . Those
who put major emphasis on te predicate (fire) feel that the anllmea is predicate
qualified by subject, that is, fre qualifed by hill or hilly fire. Those who put
equal emphasis on both subject and predicate feel that the anllmea is subject
plus predicate. Thus, we get various answers to the question as to what exactly
we infer or what precisely is te anllmea.
Pdharmati
Pak$adharmata constitutes one of the two necessary grounds for the process
of anumana , the other being vyapti , which we discus s subs equently.
Pak$adharmata stands for the relation between pak$a and hetu. It consists in the
judgmentalization of the perceptual cognition of hetl located in pak$a. It is a
relational cognition in which hetu is cognized as being a property (dharma) of
pak$a. Pak$a is said to possess two different properties, which are hetu and
sidhya. The fact ofpak$a being the sidhyadharl, that is, possessing the sadhya
as its property, is inferred on the ground of the pak$a being hetudharml, that is,
possessing the hetu as its proper. Since hetudharmata of pak$a is directly
cognized (i. e . , not iferred) , and it stiulates the process of anumana, it has
been taken to ba more significant property than that of the sadhya. Wthout te
cognition of pak$a possessing hetu as its propert, the process of anumana
cannot start or be valid. This relation of hetu and sidhya, therefore, is given
special consideration. To take the classical example of the anumana of fre on a
72 Buddhist Epistemology
hill on the basis of the perceptual cognition of smoke there, the cognition of
smoky hill is characterized as pakadhannatt. Wthout the cognition of smoky
hill , it is not possible to infer the fery hill . Mere knowledge of te vtpti between
smoke and fre cannot generate the knowledge of the fre on te hill uless the
smoke is also cognized on te hill . The relation between hill , which is paka, and
smoke, which is het, is possible only when te two are cognized to be so
related. The stda, which is known in vtpti, is a universal , but the stdha,
known through the paadhamLGtt of het, is a particular one. In oter words,
the vtpti relation is always between two uiversals, and therefore at te back of
the two relata, tere are no individuals to fall back upon, but that is not the case
with paadhamLGtt. Here, bot types of dhamLGtt of paka, whether tat of
het or stha, are establishe as having a inmlediate reference to some idividual
facts. Thus, for exalple, te vtpti sambandha between smoke and fre is
between two universals, but te pakadhannatt sambandha is grounded in the
conceptual cognition of het and paa, which also have some underlying svalak
a(LGS (svalakG(za p!$!ha bhavi) . The relation of pakadhannatt is conditional
and contingent. In tis respect it difers from te vtpti relation, which is necessary
and unconditional . The conditionality and contingency of the pakadhannatt are
because of the fact that the paka does not necessarily possess hetu as its dhanna,
as it depends on te presence of certain conditions . Sinlilarly, te het also is not
exclusively present in paae It can be present in any oter thig where conditions
conducive to its presence are available; for example, it is not necessary that the
hill should always be smoky or tat the smoke must be present only on the hill .
The relation of pak$adhamzatt is te relation between a substantive and an
adjective-dhamlz-dhana sambandha. Paka is always a substantive, and hetu
an adjective. The order of the relata can never be reversed, and if it is reversed,
then it will be an altogether different thought process. The idea of smoky hill
necessarily inlplies tat smoke is the dhanna and hill is the dhannf. If the
position of te relata is reversed, then it will be a case of " smoke on the hill , "
and in tis case hill becomes the dhanna, and smoke becomes the dhannf. The
concept of " smoky hill " is altogether different from the concept of the " smoke
on the hill . "
The henl that is a dhanna of pak$a is not any object tat may be taken to
characterize that pak$a. It should be rater such an object that is essentially in
the form of a lilga being necessarily concomitant with another object that is its
liligin. The implying power of henl is by virtue of its being a necessary associate
of stdha, and, therefore, only such a hetu is a dhanna of paka. Thus, in the
case of smoke it is in enabling us to infer fre tat it is a het, not as smoke-qua
smoke.
Vyapt or Aviabhava
The concept of vtpti is central to every theory of anumtna. Vtpti stands
for a relation of necessary concomitance, positive or negative, between two classes
of facts such that one ievitably leads to the other. The entire process of anumtna,
Te Buddhist Teory of Inference 73
as has been said, is based on such a relation between the linga (one that leads)
and te lingin (one that is led) , which can be understood in terms of necessary
dependence (avinabhtva niyama) . The significance of vapti is indicated by the
facts that there can be no anumana in its absence and tat every defmition of
anumana directly or indirectly has to bring it i.
Though we do not have any information about te analysis of vapti by pre- .
Dinnaga thinkers, Dinnaga referred to it in his statement expressing the
paladhanata and vapti as te two grouds of anumana. I te Pama{lG
Samlccaya46 he writes , " Grahya dharmastadansena vyapto hetl(l . " In
DharmakIrti we find its detailed analysis. Etymologicaly, the word "avinabhtva"
means "a + vina + bhava, " tat is " in the absence of sada necessary absence
of hetu . " Here, the term "vina" stands for absence of sadha, the term "bhava,
stands for presence of hem, and te term "a" stands for the negation (of the
presence of het) . Thus, it inplies absence of hem in the absence of sidha.
Apparently, it seems to have a negative meaning, but it has a positive inlplication
also. Avinabhava is a relation between two such tings where one thing cannot
be present without the oter. Dharmaklrti also uses te expression "svabhava
prtibandha, " which means "existential tie. " "Existentia tie" means the existence
of one thing tied to tle other, which means a dependent existence. This may be
in the form of a causal relation or an analytical relation. For example, the
dependence of effect is known to us. Sinlilarly, an analytically deduced fact by
its very essence depends on the fact from which it is deduced. Thus, there is a
svabhtva prtibandha (existential tie) between cause and effect and between the
deduced object and that from which tere is deduction. The example of the
former type is te relation between smoke and fire, and of the latter, between
rose and fower. We can deduce one fact from aother only if te two facts are
sianeous, and we can have causal connection only between two such facts
that are in il1ediate succession.
It may be asked, Why is it tat we can infer one fact from anoter only if
there is existential dependence? The answer given by Buddhist logicians is tat
tis is so because if a fact is not dependent on te oter, it cannot be invariably
and necessarily concomitant wit the latter. 47 There will be no invariability. 48
Thus, the possibility of deducirg one fact from anoter depends on an invariable
and necessary connection that precludes te existence of one witout the existence
of oter. Therefore, if two facts are existentially connected, we can assert that
one of them cannot exist independently of the oter, and therefore from the
presence of the one follows the presence of the oter.
The svabhtva prtibandha (existential tie) is always that of the linga with
lingin. Explaining tis, Dharmottara writes that linga is always dependent on the
lingin, whereas te lingin is independent of linga. That is why linga is said to be
" tied-to, " and te ling in is said to be "not-tied. " That which is tied is te gamaka,
and the one to which it is tied is said to be its gama. Tat which is not existentially
tied to anything particular will not have te relation of invariable concomitance
wit te thing, but that linga tat is not utied has the relation of invariable
74 Buddhist Epistemology
concomitance with its lilgin (which itself is not tied) because the foner does not
deviate in its occurrence with the latter. This invariable concomitance is kown
as avinabhiva. On the basis of this invariabilit the gamya-gamaka-bhiva is
established . In the absence of invariable concomitance there can be no
gama-gamka-bhiva because the Zinga by itself cannot reveal the lingin, unlike a
lamp, which can illuminate its objects. The detenination of lingin is on the
basis of invariable concomitance. Thus, on te basis of svabhava prtibandha
tere is avinabhiva, and fom avinabhiva follows gama-gamaka-bhava. This is
meant when i is said tat on te basis of svabhiva prtibandha alone one object
leads to another. 49
Hetu is known as vapa, and stdha is known as vapaka. According to
Dharmaklrti, vapti being te dhamza of both vapaka and vapa, it can be
expressed in two ways : te presence of vapaka is necessary for the presence of
vapa, and vapa can exist only when vapaka exists. These two conditions
are, respectively, known as anvaa and vatireka. 50 Arcata has explained this as
follows : An object tat has vapa as its property will also necessarily have
vapaka as its propert. This is technically known as vapakadhamzata. Here
the vapyata of the vapya is known through te vapaka. Similarly, when an
object has vtpa necessarily by virte of its possessing vapaka, ten it is
known as vtpyadharmata. By violating tis regulation there cannot be the
establishment of linga-Zingi-sallbandha. 51 Thus, in short, te avinabhiva stands
for the regulation that the Zinga can be present only in the presence of lingin and
that the absence of Zingin necessarily implies the absence of lingd.
According to Buddhist logicians, there are only two types of vapti , namely,
tadatmya and tadutatti. It can, therefore, be asked why there are only two
tpes. The answer is that the necessar connection of one with another means
that the existence of the foner is necessarily dependent on the existence of the
latter. Now one thing' s existence could necessarily depend on the existence of the
other ony under two conditions : ( 1 ) if te latter causes the former or (2) if te
latter is a part of the foner. No other condition makes the existence of one thing
necessarily dependent on the existence of another. Hence, Dharmaklrti asserts
that te relation of causality and that of essential identit are the only to necessary
relations. 52
Vapti or avinabhiva is a relation of necessary concomitance tat is universal
and invariable, but the problem is how we apprehend te vapti relation between
any two given phenomena. For tis different schools of logic in India have given
diferent answers. According to Buddhists, te relation of vapti is not directly
apprehended in perception, as all relations are conceptual construction. According
to DharmakIri, the relation of vapti is based on our kowledge of eiter causality
(tadutpatti) or identity (tadatmya) , as an effect cannot be conceived to be
independent of a cause, and hence efect i s a proof of the cause, and as regards
the two things whose natre is fndamentally identical, there can be no separation
Te Buddhist Teory of Inference 75
between the two, 53 as that would be tantamount to forfeitre of their own essential
character, which is inconceivable.
It . can be objected that Buddhists have reduced all vtpti relations to causal
or identity relations. But it can be pointed out that certain relations of uniformity
cannot be reduced to relations of causality or identity. Thus, for instance, the
impending rainfall is inferred from te movement of ants, but te concomitance
in such a case cannot be traced to causality or identity.
Buddhists may reply that mere concomitance in presence and absence cannot
constitute suficient evidence of its inevitabilit unless the contrary possibility is
debarred by the metod of tarka. The metod of tarka can be employed ony if
the relation is understood to be one of causality or identit because no oter
relation can be conceived to be invariable and uniform. It should be observed
tat tere must be a causal relation, tough indirect, between two sets of connected
phenomena. They must be coeffects of te sanle set of causes and conditions ;
otherwise, the invariability of the relation cannot be explained. Such a relation of
invariability is to be constructed or established on the basis of observation.
However, it must be noted that the mere observation of positive and
negative instances does not give rise to the knowledge of vtpti, which has to be
constructed by te conceiving mind. Of course, the mind has the capacity of
doing so, and observation only provides a stimulus for that.
In granting the constructing capacity of mind to give us the knowledge of
vapti , Dharmaklrti assumes two things : every event has a cause, and the same
cause always uniformly produces the same efect. In other words, there is general
regularity or uniformit in the universe. Whatever is a cause of particular tpe of
a thing remains forever a cause of that tpe of a thing. To think otherwise,
namely, that one type of cause does not always produce only one type of efect,
is to go against the logic based on experience. "Y" cannot be treated as an effect
of "X, " even i a single case, if all "Y" is not an effect of some "X. " It is so
because we call "X" te cause of " Y" only if "X" invariably produces "Y"
Moreover, to say tat at times "Y" is produced by "X" and at times it is produced
by something other than "X, " that is "non-X, " would mean that "Y" possesses
two contradictory natures . Agai, this would suggest tat the nature of a thing
does not depend on its cause, and to grant this suggestion would mean that the
ting comes into existence without any cause. This, in t, would make it eteral
and consequently devoid of effcienc, which is te criterion of realit. So, one
type of effect can never be regarded as being produced at times by this type of
cause and at times by tat type of cause. Wen we feel tat tere are instances of
one type of effect at times being produced by tis type of cause and at times by
that type of cause, our feeling is not justified. We commit a fall acy of
nonobservation. The two effects produced by to different tpes of causes are
not merely of one type. We are deceived by their outward similarity. A close
examination may reveal tat they, in fact, belong to two different types. 54
76 Buddhist Epi stemology
Thus, according to Buddhist logicians, what we directly perceive is nothing
but svalakaza. Perception is a series of sensations without any connection or
order in them. Only the buddhi constructs a system or order out of them, through
the instrumentalit of the two necessary relations of ttdttmya and tadutatti.
Perceptal cognitions are discrete in natre, and te vikala buddhi superimposes
order and uniformity on them. The necessit underlying the relation of tadttmya
and tadutpati is superimposed upon the perceptal cognition that provides an
occasion for the fnctioning of the vikalpa buddhi.
S
Another signifcant concept that fgures in the analysis of inferential process
i sapa. Sapaa mea tat whch i si to paa. Dhtdefes it a:
Sadhadharmasamanena samano 'rthal sapalal.
55
It is siilar to paka by vire of the common possession of the sadha,
which is the inferred propert and, therefore, metaphorically called its copartner.
In other words, all those objects that possess the property to be inferred are to be
kown as sapaka in as much as they are similar to paka in that respect. The
word "sa" is a substitte for samana. 56 It should, however, be borne in mind that
tis similarity is not absolute because the possession of sadha in sapaka is a
matter of certainty, whereas te possession of a siilar property by te paka
needs to be proved. That is why some of the tinkers belonging to the early
Nyaya school define paka as salldigdhasadhavan pakal: (i . e. , paka is that
object where the presence of stdha is suspected) and sapaa as niscitasadhavana
sapakal (i . e. , that object where the presence of stdha has been well kown) .
However, the two are similar by virtue of te fact that tey possess sadha as their
propery. This property in Buddhist thought is always a samanalakaza because
no svalakaa can ever be a property. 57 Thus, in te exple of fre beig inferred
on a hill on the basis of smoke, fre is te property tat is inferred in relation to
te hill . Therefore, hill becomes a paka, and all tose instances, like kitchen,
where fire is known to be a propert, constitute sapaka, because they are similar
to paka in cOnnly possessing the property of fre.
The paa and sapaa, even toug siilar i their possession of te comon
propert, which is sadha, are dissimilar in several other respects. Paka refers
to an object that is presently given, whereas sapaka refers to such object(s) tat
have been kown in the past . Therefore, a sapaka is always to be remembered
and not presently perceived. Second, in the case of paka it is always numerically
one, but sapaka may be more than one. However, Buddhists do not isist on
numerical multiplicit of sapakas. In the Bet Cakr /amar Diraga maitais
that there should be "at least one" instance of sapaka where hem is present
along with sadha.
The Buddhist Theory of Inference 77
Aapaa or Vpaa
That which is not similar to paka is kown as asapaka or vipaka. That is
why Dharmaklrti defines it as :
Na sapalo 'sapala(1
58
In fact, it is dissimilar to both paka and sapaa. Though te basic contention
here is to point out its dissinlilarit with sapaka, its dissimilarity from paka
also becomes evident. Asapaka is dissimilar from sapaka in that it is never a
possessor of the propert cOlOnly possessed by paka and sapaka.
Asapaka can be of te following three types :
1. Difrent fom sapala (anya);
2. Contrary to sapala (vintddha);
3 . Absence of sapaka (abhtva) .
59
Of te three, abhava is te most fndamental , and ana and vintddha cmIot
be conceived so long as there is no abhava. The conceptions of anatva and
vinlddhatva include the conception of abhtva because through the analysis of
these two conceptions the tird one is revealed. Wen te ana and vinddha are
realized, they are realized as possessig the form of abhtva of the sapaka.
Thus, abhtva is conceived as someting representing the sapakabhtva directly,
and ana and vinddha are conceived indirectly.
60
Role of Sapa and Aapa
Sapaka and asapaka are basically relational concepts. Sapaka stands for
the positive relation consisting of the presence of sadha in paka, and asapaka
stands for the negative relation indicating the absolute absence of the former in
the latter. These two become elements of the process of anumana when tey are
taken as aspects of trina lilga. Here they are concered with the regulations
of te presence and absence of linga in sapaka and asapaka, respectively. The
linga must be present apart from paka at least in one sapaka, and it must be
absolutely absent in every case of asapaka. Thus, we get te two nipas of the
linga as sapake sattvam and asapake asattvam. The concepts of sapaka and
asapaa are only bilateral as tey involve the relation between paa and sthya,
whereas the concepts of sapake sattvam and asapake asattam are trilateral as
they ivolve hetl also, along wit paka and stdha. In sapaka and asapaka
the sole consideration is the presence and absence of the sadha, whereas i
sapake sattam and asapake asatvam promienace is given to the presence
and absence of hem. In sapae sattvam and asapake asattvam the presence or
absence of hem is governed by te rule of concomitance. Therefore, we have
only these two types of regulation.
78 Buddhist Epistemology
Hetu ad It Ripas Tripya ad Pa caripy
The Buddhist theory of anllmana, as we saw earlier, is rooted in the concept
of gama-gamaka-bhava. Anllmana is that thought process in which the cognition
of the linga acts as a gamaka and leads to theinferential cognition of the ling in,
which thus is the gama. No doubt this gama-gamaka relationship canot be
established between any two cognitions ; there has to be the necessary conection
in the fonn of avinabhava. This implies that oilly a particular cognition can be a
linga for anoter specific cognition, which will be its lingin. The process of
anllmana begins with the cognition of the linga. Therefore, the linga provides
the starting point of the inferential process. It is the logical ground upon which
the subsequent cognition of lingin depends. In the Buddhist tradition linga and
lingin stand for construed facts (adhavasita) rather than real facts (grhita) . The
two constred facts are so conceived that one is taken to be a linga of the other.
But, as we said, any two such facts camlOt be arbitrarily taken to be te linga and
lingin. Therefore, it has to be seen that a particular fact is really a linga of
another fact, which is its lingin.
The establishment of te relation between linga and lingin and the cognition
of the linga constitute te basis of inferential process, and therefore they can be
taken to be premises. The transition from linga to lingin is the conclusion. If the
premises are not wellgrounded, the conclusion entailed by them also canot be
logically valid. The liriga constitutes te pivotal element i the premises. Whenever
anyting is taken to be a linga in respect of another thing taken as a lingin, there
are always two possibilities conceivable. A linga may be a genuine linga if it
really stands for its lingin. But tat linga will be termed fallacious that does not
lead to something conceived as its lirigin. In such a sitation te ling in doesn' t
have that linga as its proper linga. In fact that linga i s not at all a liriga i n respect
of tat lingin. That linga may be a linga as some other lingin than the one in
question. Thus, in the case of the present lingin, that linga is not a linga but a
lingibhasa. When a linga is employed as a ground for inferring the lingin, the
linga is known as hetl. So, if hetll is not genuine, then it is known as hetvibhtsa.
Thus, a distinction has to be drawn between hetu and hetvtbhasa or, to use a
diferent set of phraseology, between sadhetu (valid reason) and asadhetu
(fallacious reason) . In oter words, there has to be some distinctive marks of
sadhetu to diferentiate it fom asadhetu. Since the process of anumtna is grunded
in hetu, if te hetu is sadhetu, te anumina will be valid, and if it is asadhetu,
then the anumtna will be invalid. In order to have a logically valid and sound
anumina we must take care to see that hetu is sat and not asat. Different systems
of logic in India have put forth a different number of characteristics a sadhetu has
to possess. Buddhists have advocated the theory of trirpya, whereas Jainas
have reduced the three rpas to one (ekana) , and Naiyayikas have added two
more nlpas to these three, making tem fve (afcarupa) .
Te Buddist Teor of Iference 19
TH THORY OF TI(Y
According to Dilnaga and Dharmalrti, anumana stands for the cognition
of te lingin on te ground of the cognition of linga. The linga has to be
characterized by tree rpas in order to be a logically valid linga. The Buddhist
system of logic insists tat there are three ad only three essential characteristics
of a valid linga. This is so because there is only a treefold requirement of te
linga being related to ( 1 ) pak$a, (2) sapak$t, and (3) asapaka. Every linga
must possess all the three characteristics simultaneously, only then it is a
trirpalinga or trilakG1a hetl and can be made use of in the process of inference.
The doctrine tat a linga must possess three essential characteristics is known as
trirpavada or the theory of trirpalinga.
(a) Diimaga's Formulation
The frst systematic formulation of the theory of trirpya is said to be
by Dirnaga, though a reference to it is al so found in the works of
Vasubandhu. 61 Uddyotakra i his Vartika cites Dirnaga' s formulation of trirpya
as follows : "amee 't ttlye sadbhave nastitasati , " that is, existence in the
anumea (ak$a), in what is l the pak$a (sapak$a) , and nonexistence in what
is not like the pak$a (asapak$a)
.
6
2
We can render in English Dinnaga' s version
as follows :
1 . Its presence in the subject of inference.
2. Its presence i n similar instances.
3 . Its absence in dissimilar instances.
63
The first condition lays down the requirement of pak$adhannata
.
It regulates
the relation between linga and anllmea. The term anllnea has been ambiguous
and is used diferently in te Buddhist and te Nyaya traditions . In the Nyaya
tradition it stands for sadha, which is the object of inferential cognition. But
for Dirnaga and his followers it always stands for te pak$a, which is a dhannf
and of which sadha is a dhanna. Though Dilnaga doesn' t seem to be very
explicit on tis, by anlmea he meant anlmeadhamlin, tat is, paka and not
anlmeadhanna because the oter expression "tattlllya" clearly means "like
tat , " tat i, like t paka (saaka) . The second condition regulates te relation
between linga and sapak$a. It states te necessity of the presence of linga in a
sapak$a, but from its wording it is not clear whether the presence is essential or
not, nor is it clear whether the presence is in all sapak$as or in at least one only.
However, it should be understood as expressing te necessity of te presence of
liliga at least in one sapak$a. This is so because the second condition refers to
the positive concomitance between het and sadha, the reference to one instance
of which is logically sufficient to flfill the requirements of an udahar{lG. The
third condition regulates the relation of linga with asapak$a and states its absence
in the sanle. Like the frst two conditions this also is anlbiguous and may be
understood as requiring te necessary absence of hetu i asaaka or as requiring
80 Buddhist Epistemology
the necessary absence of hetu alone i asapaka. The intention of Diiaga seenis
to be that hetu should be absent in all asapakas. This is because of the
requirements of inverted concomitance. Not only must there be a statement
giving out the conditions for te presence of hetll, but tere should also be a
statem.ent of the conditions tat necessitate the absence of hetl. The conditions
that necessitate the absence should be more than the conditions that guarantee
te presence of hetl , because when we talk of presence, it is always in respect of
one individual, whereas when we talk of absence, it is necessarily in respect of
te entire class. Thus, the statement of sapaka is particular in natre and that of
asapaka is universal . The element of necessity or unexceptionality is available
only in te latter. From tis it is quite evident that the logical status of the two
statements is not the same, and hence te asapaka cannot be regarded as just
the negative corollary of the first.
(b) Uddyotakara's Objection
It seems that Uddyotakara was not flly aware of te precise sense in
which Diaga talked of the three rupas of hetu. Because of the apparent
ambiguit he pointed out alternative iterpretations of each of the tree nlpas
and tried to expose their untenability on some logical grounds . Uddyotakara' s
examination of Diiaga' s formulation is as follows.
In the statement of the first nlpa, the expression "anumee sadblv
a
" was
not free from ambiguit, and as a consequence of tat, Uddyotakara examined its
two possible iterpretations, namely,
1. The hetll is present i n the paa only, and
2. The hetu alone i s present i n the paa.
To point out te difference in the two interpretations te following symbolic
formulation may be helpfl :
(i) b is present in a only every b is a no non-a is b.
(ii) only b i s present i n a every a i s b no non-b i s a.
From the preceding i t i s quite evident that these two interpretations
don' t mean the same ting, and therefore the formulation of Diiaga was
replete with ambiguit. The formulation of the second nlpa given by Diiaga
was also ambiguous in te same way, which led Uddyotakara to give two different
interpretations, namely,
(i) The hetu is present in similar instances only.
(ii) The hetll is present in "all similar instances. "
In the same way Diaga's formulation of the third Tpa was subjected to
two diferent iterpretations by Uddyotakara as follows :
Te Buddist Theory of Iference
(i) Only the hetu i s absent in the "dissimilar instances. "
(ii) The hetl is absent in the dissimilar instances only.
81
Though Diiaga has given his unambiguous version in the Hetu Cakr
Damant, where the presence or the absence of hetu in relation to sadha is
well regulated, perhaps he did not do so in te Prmala Samuccaya, from
which Uddyotakara has quoted him.
.
(c) Dharmakirti's Reformulation
Diimaga' s account of trin1pya was given a more rigorous form by
Dharmaklrti . He did so by associatig the word "eva" wit each of the tree
marks and by qualifig the entire expression by te word "niscitareva. " The
other modification he introduced was the use of the terms "sapaka" and
"asapaka" to remove all sorts of ambiguities.
We find that it was only to give Diiaga' s formulation more precision that
Dharmaklrti put te emphasizing particle "eva" with te entire expression. Wth
tis addition the frst condition, for example, gives te emphatic meaning of
"impossibilit of absence" to the assertion.
(d) Dharmottara's Modifcations
The ambiguity of nonquantification tat vitiated Dinnaga' s formulation
was sought to be removed by Dharmaklrti by adding the restrictive word "eva,
but unexpectedly, his modification caused frther ambiguity. That is why
comentators fom Dharmottara onward had to struggle hard to interpret it in a
satisfactory way. Dharmottara had to face a dilemma on tis account while
commenting upon the second and third mpas . The second nlpa apparently meant
"only i the sapaka hetu lust be present, " but b implication it meant the hetu
must be absent in all asapakas. But then the second n1pa will be another way of
saying the sane thing as te third mpa. This renders the third n1pa superfluous,
but this was not the intention either of Diraga or of Dharmaklrti.
In order to remove this ambiguity Dharmottara suggested another
modification, that the expression "niscitam" should be added to all the three
nlpas to rendcr them free from ambiguit. 6
Thus, his statement, " lilJgasyanumee eva sattvam sapake eva sattvam
asapake casattvameva niscitam" can be interpreted as follows :
1 . Li1gasyanwnee (ea) sattam nifcitam.
2. Sapae (ea) sattam nifcitam.
3 . Asapake casattam (ea) nifcitam.
In English it can be rendered as follows :
1. Existence only (never nonexistence) i n the paa.
2. Existence only i n those things that are similar t o paka (never i n things
that are not similar to paa) .
82
Buddhist Epistemology
3 . Only nonexistence (never existence) in things tat are not similar to pala.
(e) Stcherbatsky's Iterpretation
Stcherbatsky has expressed Dharmaklrti' s version in three diferent ways as
follows :
(i) The presence of the reason in the subject, its presence "just, " that is, never absence.
(ii) Its presence in similar instances, "just" in similars, tat is, never in dissimilars, but
not in the totality of the similars.
(iii) Its absence from di ssi mi lar instances, its absence "just , " that i s , never
presence, absence from the totality of the dissimilar instances.
It seems that the expression of the second condition just given has been
defective in two ways. First, the implication, namely, "never in dissimilars " is
not intended by Dharmaklrti, nor does it follow fom the expression sapake eva
satvam (niscitam) " as implying "necessary presence of the hetu i at least one
similar instance. " Second, the expression "but not in the totality of the similars "
is also fault insofar as it implies that there should not be presence of hetu in the
totality of similars. Its proper wording should have been "but not necessarily in
te totalit of siilars. " Because of these defects the second and te third conditions
have been seen to be mutally implyig each other.
65
The second mode of the expression of trirupya is as follows :
(i) The necessary presence of te reason in subject's totality.
(ii) Its necessary presence in similars only, although not in their totality.
(iii) Its necessary absence fom dissimilars in teir totality.
This way of expressing trirpa is free from the defects referred to earlier
because it draws a clear distinction between "presence ony i sapakas" and
"necessary presence i at least one sapaka. ' Ony when te second rpa is
uderstood in te former sense does it overlap wit te tird na, but not when
taken in te latter sense.
The tird way of expressing trirua is noting but an abbreviation of
te second. It is as follows :
1 . In subject wholly.
2. In similar only.
3. In dissimilar never.
(0 Chi's Exposition
Chi has given the Chinese version of trainipa, which seems to be free from
ambiguity as follows :
(i) The "pervasive presence" of the hetzt in the subject.
(ii) The " necessar presence" of the hetu i some similar instances.
The Buddhist Theory ofInference 83
(iii) The "perasive absence " of the hetu from dissimilar instances.
In te 'preceding rendering the use of te word "pervasive" in te first
and the third rpas, and of te word "necessary" in te second rupa refer to the
distributive quantification of hetu. Sinlilarly the phrase "necessary presence"
means "assured presence" or "not failing t be present . " Chi has defned te
notion of "pervasive presence" as follows :
"Pervasive presence of b in a. "
"b is present in every a. "
" every a is b. "
The notion of "pervasive absence" is defined by him as follows :
"Pervasive absence of b fm a. "
" b is absent from every a . "
"no a is b. "
The notion of "necessary presence" is defned by hinl as follows :
" Necessary presence of b in a. "
"b is present in atleast one a, almost every a"
"at least one a, at most every a i s b. "
In accordance with this understanding he interprets trirpa as follows :
I ri pa:
"The propery of g is present in every tg which possesses te propery J "
"Everyting which possesses the propert fpossesses te property g. "
= "For every x, x is an f implies x is g.
"
I rupa:
"There is at least one occasion in which the property g is present in a ting
which possesses the property h apart from the ming which possesses me
propert fwhich remains to be proved. "
"Apart from te thing which possesses the property f at least one ting
which possesses me property h possesses te property g.
"
"For some X which is not f, X is bom an h and g.
"
Irupa:
"There is no occasion in which me property g is present in migs which
possess me propery of non-h. "
"Nothing which possesses me propery non-h possesses me property g.
"
"For no X, X is bom non-h and a g.
"
84 Buddhist Epistemology
On the basis of this interpretation Chi proposes to resolve Dharmaklrti ' s
dilemma on the interpretation of "eva" (only) i the second nlpa as follows :
" Only presence. "
" Only presence of the property of the hetu in similar instances but not
otherwise. "
" Only presence of the property of the hetu in similar instances but not
utter absence of its property in all similar instances. "
"The presenc of the property of the hetu in at least one similar instance. "
Understood in tis way, the second rpa doesn' t overlap with the third rpa,
and the problem of redundanc is overcome.
Afer discussing the Buddhist theory of hetu-trainlpa in detail , let us attempt
a comparative account of the different theories of hetu-rpa attempted in the
three systems of logic, namely, Buddhist, Nyaya, and Jaina. All three systems
have diferences with regard to the required rpas of het. Buddhists insist on
three nlpas only, but Naiyayikas accept five nlpas of hetu, and Jainas emphasize .
that only one Tpa is necessary.
Nyaya Teory of Paicaripya
Nyaya logicians emphasize the fact that het or lilga is the central element
in te inferential process. It is related to paka as well as sadha, and it occurs in
all significant premises, namely, hetu, udaharrza, and upanaa. The validity of
inference depends, to a great extent, on hetu; and that is why a fllacious iference
is kown as having hetvabhasa. Nyaya logicians therefore have taken great pains
to expound the characteristics of a good hetu, because if the het is sadhenl, ten
the inference will be valid, and if it is asadhetu, te inference will be invalid.
Naiyayikas enumerate the following conditions of a sadhet:
1 . Padatva. Het must be a dhamza or characteristic of paka.
This means that the het should have its locus (adhikarza) i paka, that is, it
should reside in, or be related to, paka. Thus, the paka should never be without
hetu, and nothing can be paka if there is no hetu related to it.
2. Sapesattvam. Hetu must be distributively related to stdha. This
means tat whenever hetu exists, in all such cases stdha must be present. This
means that the necessary condition for the occurrence of henl is the occurrence
of sadha.
3 . VdvavtI. Henl must be absent in all such cases where sadha
is absent . This means that the absence of stdha necessarily implies the absence
of hetu. So, in all asapakas, that is, where stdha is absent, het must also be
absen.
4. Abadtavi$ayatvam. Hetu should not be such as to be contradictory to,
or icompatible with, te nature of stdha. This is so because te fnction of
het is to establish sadha, but if it is incompatible with sadha, it will disprove
stdhya.
Te Buddhist Teory ofInference 85
5. Asatpratpaatva or Aviuddhatva. Hetu should not be such as to
have a rival hetu that is competent to invalidate it.
Of cOlrse, Naiyayikas make it clear that all these five Tupas are not
simultaneously present in all hetu. For example, in a kevaltnvayi anumtna
the third condition will not be applicable. Similarly, in the kevalavatiTeki
anumtna te second condition will not be applicable. 66 Hence, it has been
said that a saetu satisfes te five or at least the four conditions.
If we compare te Nyaya theory of pafcaTUpya with the Buddhist theory
of triTUa, we find that the frst three TUpas of the Naiyayias correspond
to te triTUpya of the Buddhists. Of course, the details in the formulation of
these three TUpas are very much diferent in the two traditions, but we can still
regard them as having some rough correspondence. But the last two TUpas of the
Nyaya theory are not accepted and will not be acceptable to Buddhists. Now it
can be asked why Buddhists do not accept the last two TUpas. The reason is that
Buddhists believe in the theory of prmtvavastht, which is the opposite of
the Nyaya theory of prmtlasmipalava. So the verdict of one anumtna, if
falsely drawn, can be contradicted by some other prmtla in the Nyaya position,
but the verdict of one anllmtna carmot be contradicted by another prmtlJa in
the Buddhist position. Therefore, there is no possibility of satrtipaka here. 67
Second, the basis of anumtna, according to Buddhists, is liJiga-lingl-bhtva,
between any two concepts. We shall never establish lilga-lingl-bhva if they are
of oppos ite nature . Therefore, there wi l l never be an occas i on for
abtdhitaviayatvam.
Third, the introduction of te particle eva in (2) and (3) at proper places
implies that a hetll has a necessary conection with the stdhya. Thus, as a
hetl 's necessary conection with the stdhya is covered by conditions (2)
and (3)-they being a modification of the corresponding conditions of the Nyaya
Vaiseia scheme-tere remais no possibility of its being either counterbalanced
by another hetll or contradicted by a noninferential prmtlJ a. So, for Buddhists
conditions (4) and (5) formulated by the Nyaya-Vaiseika logicians become
redundant or inapplicable.
Though tere has been a lot of controversy beteen Buddists and Naiyayikas
as to whether only three TUpas or five TUpas are to be accepted, te whole
controversy appears t be ftile because they are two diferent traditions having
different tpes of requirements. The Nyaya tradition stands in need of five n7pas,
whereas in te Buddhist tradition only three Tupas are possible, and terefore the
teories of pafcanlpya hetll and trinlpya hetll are perfectly in order in their
respective systems.
J aina Criticism of Taiipya
The Jaina tradition maintains that hem has a unitary nature, and therefore it
possesses only one na. Hem is defined as tat which is iseparably connected
with stdha. This means hetu is tat which is present ony i the presence of
stdha and which is necessarily absent i the absence of stdha. AkalaIdeva
86 Buddhist Epistemology
defines hetu as "sadhyavinabhtva vinibodhaikalaka1a
.
" 68
Some Jaina logicians
have described tis fact with the help of aoter concet, namely, "anaht
n
upaat.
Vadidesii puts it as "niscitanathtnupapatteka lakalOhetub. " 69 The teml
" anathanupapatti " means tat in the absence of sadha (anatha) , there should
be absence of hetu (anupapattO. By implication it means that hetu should be
present ony in the presence of sada. So to be sadhavinabhtvi is the sole
characteristic of hetu.
The Jaias have not ony insisted on one and ony one characteristic of
hetl but they have also put for a reftation of the theories of trinlpa and
pafcarpya.
Patrasvamin seems to be te first Jaina scholar, as quoted by Akalalka
and VinItadeva, who attempted reftation of the doctrine of trirpa as follows :
Anyathtnupapannatvam yatra tatTa trayeIa kim
Ntnathtnupapannatvam yatr tatr tryeIa kim, 70
In the preceding verse Patrasvamin emphatically maintains that
"anathtnllpapannatvam" i s the only rpa of het. If i t i s present, then what is
te need of three nipas, and if it is absent, ten what is the purpose of having
three rpas? Following Patrasvamin, all subsequent Jaina logicians lie AalaIa,
Vidyananda, Prabhacandra, Vadidevasuri, Anantavlrya, and Hemacandra have
also refted the doctrine of trirpya. The preceding verse of Patrasvamin has
been quoted by Santarakita in the Tatta Sgrha in the form of parva-paka
presenting the Jaina position, and there he has refted the Jaina objections against
trirpya. The basic objection put forth by the Jainas is that even an invalid hetll
may have the tree rpas, "tasa hetvabhtsasyapi sambhavat. " For example, the
argument "There is smoke on the hill because there is fre" has all the three
nlpas, namely, pakadhamzatvam, sapakesattvam, and vipakeasattvam, but
still it is logically invalid because tere is no necessary connection (avinabhava)
between fire and smoke. In oter words, there may be fire and smoke on the hill ,
and so the condition of pakadhannatvam is flflled. Similarly, there can be fire
and smoke in the kitchen, a te condition of sapkesatam is flflled. Liewise,
there will be no fire and smoke in the lake, and the condition of vipakeasattvam
is also flfilled. Even then fire is an invalid hetll for smoke. Of course, Buddhists
may point out that the trirpya should be understood as qualifed by avinabhtva,
and there is no avinabhava between fire and smoke because there may be fire
without smoke, for example, in an electric hearth. So the relation of fire and
smoke is conditional (sopadhika) and not unconditional (nirupadhika) . To this
Jainas have replied that if avinabhava is to be emphasized, then this amounts to
their position, which means hetu has only one rupa and not three rpas. Jaina' s
point of emphasis is that te hetu in an argument is valid and competent to prove
the sadha not because the two exist in the same paka, as te Naiyayikas say,
but because the to are inseparably related. In the words of Jaina logicians the
hetu is not cognized oterwise than in conection wit te sadha. Accordingly,
The Buddhist Theory ofInference 87
so far as the essential characteristic of valid hetu is concered, it is this and this
one only.
Santarakita has stated and exmined Patrasvami' s arguments in great detail .
The basic point made by Patrasvamin is that hetu is valid if and only if it has
"anyathinupapatti " with respect to sadhya. This is the only featre of hetu that
is required, and it takes care of pakadharmatvam, sapakesattvam, and
vipakeasattvam. To this Santarakita replies tat this concept is nothing but an
abbreviated form of trirupya. Anaythinupapatti includes positive and negative
concomitnce, and by its presence in the object that is paka, pakadharata is
also admitted. Santarakita not only has defended trirupyavada against
Patrasvamin' s criticism but has himself raised some obj ections against
Patrasvamin' s theory of ekarupya.
71
Helu Car Qaaru of Diimaga
The doctrine of triripya has been exlicated by Diiaga in his work Hew
Cakr pamaru,
7
2
a primer of Buddhist formal logic. In this work nine diferent
conceivable relations of hetu with paka, sapaka, and asapaka are presented.
This presentation of the doctrine of trirapya in a wider context was named by
him as the doctrine of hetucakr.
73
Both these doctrines refer to the extension of
hetu. This doctrine of trirpa represents only three valid relations of hetll with
paka, sapaka, and asapaka, whereas the doctrie of hetu cakr takes for
granted the presence of hetll in paka, and its relations with sapaka and
asapaka are alone taken into consideration. Here three possible ways of the
relation between letu and sapaka and letu and asapaka are conceived and
analyzed, namely, vapaka (pervasive presence) , avr i (pervasive absence) , and
ekadeiavr i (partial presence) . 74 He gives a formal schema of nine valid and
invalid types of anumana based on three possible relations of letu with sapaka
and asapaka. Since the text is very short we can have the advantage of givig it
here in fll . 75
.
Helu CaJaaru (Te Weel of Reasons)
Homage to Mafjusrlkuarabhita.
Homage to the Omniscient One, who is
The destroyer of te snare of ignorance.
I am expounding the determination of
The reason with three-fold characteristics
Among the tree possible cases of "presence, " "absence" and "bot"
Of the reason in the probandum,
Only te case of its "presence" is valid,
While its "absence" is not.
The case of "both presence ad absence" is inconclusive,
It is therefore not valid either.
The "presence, " "absence" and "both. "
88
Of the reason in similar instances,
Combined with those, in dissimilar instances,
There are tree combinations in each of three,
The top and the bottom are valid,
The two sides are contradictory.
Buddhist Epistemology
The four corners are inconclusive because they are "too broad, "
The Centre is inconclusive because it is "too narrow. "
Knowable, produced, non eteral,
Produced, audible, efort-made,
Eteral, efort-made and incororeal,
Are used to prove the properties of being:
Eteral, non eteral , efrt made,
Eteral, eternal, eternal,
Non effort-made, non eternal and eternal .
When two tops or two bottoms meet,
The reason is valid.
When two corresponding sides meet,
It is contradictory.
When corresponding corners meet,
It is inconclusive because it is "too broad. "
When the centers of two crosses meet,
It is inconclusive because it is "too narrow. "
Since there are nine classes of reason,
Accordingly we have nine sets of example:
Space-pot, pot-space,
Pot-lightning-space,
Space-pot -lightning,
Lighting-space-pot,
Pot-lightning-space,
Space-action-pot.
The above concerns the "determined reason
"
onl y;
A regards the "doubtfl" ones,
There are also nine combinations of
" Presence, " "absence" and "both. "
Here te presence of hetli paka i s combined with diferent types of relations
that a hetl may possess with sapaka and asapaka. The combination of these
two sets of possibilities gives rise to nine tpes. Among tese nine possible
relations between hetl and stdhya, only two are valid, and tese constitte the II
and III nlpas of the doctrine of trirpya, and the remaining seven are graded as
invalid as they are at variance wit the doctrine of trirpa. The nine possible
The Buddhist Theor ofInference
relations are as follows :
89
I : Sapalasapalavapaka: Pervasive presence i n similar and dissimilar instances.
II: Sapalavapak asapalav.rt i : Pervasive presence in similar instances ad necessary
absence in dissimilar instances.
II: Sapalavapaka asapalaikdeiv.rt i : Pervasive presence in similar instances and
partial presence in dissimilar instnces.
IV: Sapalavrt asapalavapak: Necessary absence i similar instances and pervasive
presence in dissimilar instances.
V: Sapalasapalavrt i : Necessary absence i n similar and dissimilar instances.
VI: Sapalav
.
rt asapalaikdeiavrt: Necessary absence i similar and partial presence
i dissimilar inst
a
ces.
VII: Sapalaikdeiav.rtti asapalavapak: Partial presence in similar instances and
pervasive presence in dissimilar instances.
VIII: Sapalaikdeiavrt i aapakavrt: Partial presence in similar instances and necessary
absence in dissimilar instances.
IX: Sapalasapalaikadeiavrti : Partial presence in similar and dissimilar instances.
These nine rupas of the hetu cakr have been represented by Dil naga in the
form of tis diagram:
I
.
IV
VII
II
V
VIII
Referring to this diagram, he writes :
The top and the bottom are valid,
The two sides are contradictory,
III
VI
IX
The four corners are inconclusive because they are "too broad. "
The center is inconclusive because it is " too narrw. "
Keeping this in view, we can explicate te diagral as follows :
I (corner)
IV (side)
VII (corner)
II (top)
V (center)
VIII (bottom)
III (corner)
VI ( s i de)
I X (corner)
Afer stating the hetu cakr, Dilaga proceeds to illustrate the different
nlpas with the help of the following examples :
Knowable Produced Noneteral
Produced Audible Efort-made
Eteral Efort-made Incororeal
Eternal Noneternal Efrt-made
Eteral Eteral Eteral
Non effort-made Non eternal Eternal
90
Buddhist Epistemology
Wth te help of these examples we can explai te earlier mentioned verse
like this :
1o tops meet:
Whatever is produced is noneteral :
1o Bottoms meet:
Whatever is efort-made is noneteral:
1o corresponding sides meet:
Whatever is produced is eteral :
Whatever is efor-made is eteral :
Corresponding corner meet:
Whatever is knowable is eteral :
Whatever is eteral is nonefort-made:
Whatever i s noneternal i s effort-made:
Whatever i s incorporeal is eteral :
To Centers meet:
Watever is audible is eteral:
VALID.
VALID.
Contradictory.
Contradictory.
Inconclusive, " too broad. "
Inconclusive, " too broad . "
Inconclusive, "too broad. "
Inconclusive, " too broad. "
Inconclusive, " too narrow. "
The statement of te hetu cakr, because of its cryptic worg, is unntelligible
by itself, but with the help of examples given by Diraga it can be better
uderstood. Here we take a example of anumana consistig of tree steps because
Buddhists accept only tree steps. The tree steps are prtjna, hetu, a udaharla
of sapaka and asapaka. The three terms of anumana are paka, het, and
sadha. It is presumed tat hetu is present in paka (i . e. , pakadhamzatva) . The
relation of het with sapaka is conceived to be of tree possible types, namely,
vapaka, av
.
rt i, and ekadeiavrti. Siilarly, te relation of hetu with asapaka
can also be conceived of i te precedig maner. On te basis of the preceding
stipulations te hetu cakra can be exemplifed i te followig way:
Statement of the eple of frst form:
Prtjii Sound is eteral.
Hetlt because it is knowable.
Udaharla (of sapaka and asapaka) : like space and unlike pot.
The three terms are:
Pala sound.
Hetll knowable.
Sthya eteral .
In tis example te hetu, apart fom being present in paka, is also present
i bot sapaka and asapaka. It renders this argument invalid. Here tere is a
fallacy of sadharza anekantika (inconclusive, too broad) , because the hetu is
present in all three, whereas as per rule it should be present only in te first two.
Te Buddist Teory of Iference
Statement of the example of second form:
Prti Sound is noneternal .
Hetll ' because it is produced.
Udaharala (of sapaa and asapaa) : like a pot and unlike space.
The three tenus are :
Paka sound.
Hem produced.
Sadha noneteral .
91
In this example the hetti s present in paka, i s also present in sapaka, and
is absent in asapaka. Thus, it satisfies all te three requirements of trirCpa.
So it is valid.
Statement of the exmple of third form:
Prtina Sound is efort-made.
Hetl because it is noneteral.
Udaharala (of sapaka and asapaka) : like pot and unlike lighting and
space.
The three ters are:
Paka sound.
Hetl
Sadhya
noneteral.
effort-made.
In this exmple, the hem is no doubt present i paka and sapaka but is
absent in some instances of asapaka and is present in others. The hetl has
asapakaikadesavrtti because ( 1) some noneffort-made things are noneternal
like lightning whereas (2) some nonefort-made things are eteral like ether.
Thus, the presence of hetl in some asapakas renders this argument invalid.
Here there is a fallac of sadharla anekantika (inconclusive, too broad) .
Statement of the example of fourth form:
Prtina Sound is eteral .
Hem because it is produced.
Udaharla (of sapaka and asapaka) : like ether and unlike pot .
The three terms are:
Paka sound.
Hettt produced.
Sadha eternal .
In this example, the hem is present in paka, is absent in sapaa and present
in asapaka. Thus, hem is contradictory of sadha. This renders it invalid. Here
there is a fallac of viruddha (contradictory) ,
Statement of the example of ffh form:
Prtina Sound is eteral .
Hetu because it is audible.
92 Buddist Epistemology
Udtharal;a (of sapaa and asapaa) : Like space and unlie pot.
The three terms are:
Pak$a soud.
Het audible.
Stdha eteral .
Here te hetu is present only in pak$a and is absent not only in asapak$a but
also in sapaa. Here thereis a fallac of astdhtrza anektntika (inconclusive,
too narrow) .
Statement of the example of sixth form:
Prati ft Sound is eteral.
Hetu because it is effort made.
Udthar1a (of sapak$a and asapak$a) : Like space and unlike pot and
lightning.
The three terms are:
Pak$a sound.
Het effort made.
Stdhya eteral .
Here the het is present only in pak$a, absent in sapak$a, but present in
some instaces of asapaa and absent i others. Here there is a fllac of vimddla
(contradictory) .
Statement of the example of seventh form:
Patift Soud is nonefortmade.
Hetu because it is noneteral .
Udtharla (of sapak$a and asapaa) : Like lightning and space and unlike
pot.
The three terms are:
Pak$a sound . .
Hetu noneteral .
Stdhya noneffort made.
Here the hetu apart fom its presence in pak$a is present in some instances
of sapak$a and absent in others, whereas it is present in asapak$a. Here there is
a fallac of stdhtrlJa anektntika (inconclusive, too broad) .
Statement of the example of eighth form:
Prtijft Sound is noneteral .
Hetu because it is efort made.
Udtharza (of sapak$a and asapaa) : Like pot and lightnig and ulike
space.
The tree terms are:
Pak$a soud.
Hetu efort made.
Te Buddhist Teory of Inference
Stdhya noneteral .
93
Here the hetu is present in pak$a, present in some instances of sapak$a and
absent in others, whereas it is absent in all the cases of asapak$a. Thus, the
argument satisfies all the three requirements of the trirupya. So, it is valid.
Statement of the example of ninth form:
Prtifi : Soud is eteral .
Hem because it is incorporeal .
Udahar{!a (of sapak$a and asapak$a) : like space and infinitesinlal particles
and ulike action and pot.
The three terms are:
Pak$a soud.
Hetl icorporeal .
Sidha eternal .
Here, no doubt, hettl is present in pak$a and i some instances of sapak$a
but is also present in some instances of asapak$a. This renders it iconclusive
(too broad) . Here there is a fallac of stdhir1a anekantika.
T AlYAlS
Anumina is a complex cognition arrived at as an outcome of a thought
process consisting of several stages known as avayavas. The entie inferential
thought process, when expressed i language, is said to constitte one single unit
named as vika, having several constituent elements (avayavas) . 76 In case of
prtak$a being a simple cognition (in the Buddhist tradition) there are no parts,
but that is not the case with anumina, which is a complex cognition having
several parts closely connected together by mutual requirement (arspar
apek$ita sambandha) .
DaavayaviTadition
As regards the complex character of anumina, tere is no difference of
opiion among the diferent schools of Indian logic. But how many parts, tis
complex consists of, has been a matter of dispute. Several positions are available
in India logical tradition about te nulber of avaavas rangig from ten to one.
Vatsyayana77 refers to a view that held that avaavas of anumina were ten,
namely, jifisa, Sa1lsaa, saka pripti, praojana, sa7!lsaavudasa, prtifi,
hem, udahar{!a, upanaa, and nigamana. It seems to be a reference to some
Jaia logicians of early tinle. Bhadrabahu, for example, refers to te dasivaavi
vika. He is of te view that te nuber of avaavas is really two only, namely,
prtifi and hem, but it may be increased up to ten, depending on the grasping
power of the other party. However, it is to be noted that Vadideva even grants that
in dealing with a particular type of person, even one single avaava, namely,
hetti, may constitute anumana. Those who insist on ten avaavas were of te
opinion that the five avaavas have to be added, taking psychological factors in
consideration. But Vatsyayana and oter Naiyayikas criticize tis view on the
94 Buddhist Epistemology
ground that the frst fve members represent not so much the logical grounds
needed for drawing the conclusion as the psychological or epistemological
conditions involved i anllmana. Jinisa, tough psychologically an antecedent
condition of discussion of hetu, has no bearing on the hetu and so is outside the
limit of hetu, which aims at provingprtii. Sallsaa, too, is a purely psychical
fact, and though it inspires the inquiry and as such initiates te argument, it has
no probative value.
S
akapripti likewise is a question of psychological attitude
and by itself has no logical consequence. Similarly, praojana is only a by
product of the syllogistic argument isofar as it stands for practical application of
the truth ascertained. The fifh factor, namely, sGJlsaavldasa, which is arrived
at by the method of tarka, has a remote logical bearing on doubt, but it cannot,
for that matter, be regarded as a part of the process of anumina because its
fnction is only to approve and confirm the truth of the conclusion logically
deduced fom the premises. Thus, te Naiyayikas point out that the first five
avaavas are of no logical use and therefore do not constitute a part of the
process of anumana. But the remaining fve avaavas from prtina onward are
the real avayavas, as each of them is essential to te deduction of the conclusion
and to te conviction of the disputant. None of them can be omitted, as each has
a distict contribution that is not made by the other.78
Naiyayiks consider the last five to be the real avaavas, and therefore teir
model is kown as pancavaavivaka.
79
In this context one may refer to a very
interesting remark, though not very plausible, made by Vatsyayana that in the
Nyaya school only five avaavas have been accepted owing to a desire to equate
"four premises " with the four prmazas: prtina sabda, het anumana,
d!$!anta (udaharla) prtak$a, and llpanaa llpamana.
80
It is noteworthy
tat anllmana is here identifed with hetu only. The Vedantins admitted only the
frst or the last three members of the pancavaavi vaka.
81
A similar view is
expressed by the Mlmamsakas.
Pa civayavi'adition
According to Naiyayikas, prtina is the first avayava because it is the
statement of the thesis to be proved in the process of anumana. It is the first
avayava insofar as it is the starting point of the process of anumana. The second
avayava is hetu, which consists in adducing a fact capable of proving te siJa.
The third avayava i s udaharza, which stands for a familiar instance known to
possess te hetu, which is invariably concomitant wit its stdha. The fourt
avayava is upanaya. It is an application of hetu to te disputed case after its
probative value has been attested by the udaharra. In the tenninology of later
N aiyayikas, it is a statement of paramarsa.
8
2
The last avayava, according to
Naiyayikas, is nigamana, which is a restatement of te tesis tat is well established
through hetu and udaharrz a. The Nyaya pancavayavi vaka can be illustrated
with the help of te classical example:
Prtina Word is noneternal .
Te Buddist Teory of Inference 95
because it has a definite origin. Het
Udahar/a A thing having a definite origin is seen to be
noneternal , as, for instnce, a pot .
Upanaya
Nigamana
Word has a defmite origin.
Therefore, word is noneteral .
Rejection of Pa cavvavi Tdition
In te Buddhist tradition there is a general acceptance of three avaavas,
which may be either te first three or te last three. 83 The Buddhist formulation
of anllmana can tus be:
Word is noneteral,
because it has a defite origin.
A thing having a defmite origin is s t b nonetr, a, fr instce, a pot.
Dharmaklrti is perhaps te frst Buddhist philosopher who questions the
Nyaya theory of pafcavaavl vaka. He argues that prtijfa has no probative
value and, therefore, is not a part of te process of anlmana. Ngamana is
proved by te vire of te second and te third or te third and te fourh
members, te fourt and the second members being identical i import. These
two members alone are relevant, and since te nigamana follows irresistibly
from tese two, it need not be taken to be a part of te process of te anlmana.
Moreover, nigamana is only a useless repetition of prtifa and so should not be
taken to be an avaava.
Vacaspati and Jayanta have, on the other hand, put fort a strong defense of
. pal1cavava. Vacaspati argues tat te arguer (vadin) has to state what is wanted
by the opponent; otherwise, his or her statement will fail to receive atentive
consideration at te hands of te opponent. The pratijfa, that is, " Word is
noneteral, " is what is wanted to be proved by both te opponent and the arguer,
and unless tis is stated, the hetll, "because it is a product, " will be completely
irrelevant. He, of course, concedes that the statement of prtijfa has psychological
value as it directly enlists te attention of the opponent to the subject matter of
dispute and tus renders the whole argument effective. The Buddhist model ,
of course, lacks this initial advantage and may fail to rouse the attention of te
opponent.
The defense of Vacaspati does not seem to be convincing enough. The arguer
advances an argument only when there is an occasion for it, and te opponent,
too, canot but give attention to the argument if he or she is serious of te
purose; oterwise, te Naiyayika' s prtijfa, too, would not receive atention
from the opponent. 84
Jayanta defends pafcavayavl on a different line. He contends tat the arguer
should try to brig home an argument exactly in the order in which he or she has
hin1self or herself reasoned out the conclusion. The arguer first observes the
pak$a beig associated wit hetu and ten remembers'te universal concomitance
96 Buddhist Epistemology
between hetl and sidha. Although the paka is observed witout te sadha,
te association of sadha with te paka is stated in te prtijia on te ground of
the observation of hem, but nonobservation of such an association leads to a
feeling of ucertint for which a proof has to be requisitioned. Jayant' s position
also seems to be quite weak even from te psychological point of view because
only in a complex and complicated situation do we need all tree avaavas;
otherwise, i normal situations hem (reason) is enough.
Buddhists terefore contend tat in an anumana te avaavas should be
chosen keeping in view teir logical cogency, and the psychological order
should have no bearing on it . If te psychological factors have a determining
val ue , then even such factors as s tatement of approval shoul d b e
incorporated as an avaava. If logical cogency i s regarded as a criterion,
then prtjia should be excluded i te same way in which jiiasa, samsaa,
and so on, advocated by the exponent of dasavava, have been discarded
by Naiyayikas .
According to Buddhists, upanaya, which is an application of hetu and
udaharza, too, is superfuous and has no independent probative force, as it
only reiterates the hem and udahara.
85
To tis objection of Diaga and Dharmaklrti , Uddyotakara has given the
following reply. 86 The statement of hem serves only to assign a reason and doesn' t
testif tat te hem exists i n the paka. The lpanaa emphasizes the presence of
hem in te paka and as such has a diferent fnction and value fom te mere
statement of hem. Furer, the upanaa doesn' t reiterate te hem as such; it
applies te hem along with vapti and udahara to the case in question and thus
leads to te conclusion. The statement of hem unbacked by vapti is ineffective
and hence needs upanaa.
Santarakita, a later exponent of te Diaga-DharmakIrti tradition reacting
to the Nyaya contention, argues that if lpanaa is deemed necessary for pointing
out the presence of hetl in paka, then this amounts to usurping the task of hem.
In that case what shall be te fnction of hem? Mere sttement of hetl witout
rfernce t it prnce ipaa is absolutely out of place. Therefore, a meaningfl
statement of hem should always be in relation to paka and sadha, and if these
two fnctions are served by the second and the tird avaavas, then what is the
need of having lpanaa? In fact, the statement of Jdhamzata is the basis of
anlmana, and if this is done by bot te second and the fourth avaavas, then
eiter of them should be jetisoned, and lpanaa, is not necessary for bringing.
into relation the vapti wit hem, because tis relation is understood eo ipso
from the two premises, namely, hem and ldaharza, and tese two premises
constitte a perfect anlmana, and noting frter is necessary.
In Srldhara we fmd a reply to Santarakit regarding te fnctions of hem
and lpanaa. According to hi, te two don' t serve te same fnction. The
second avayava, that is, hem, also kown as apadeta in te Vaiseia tradition,
is the bare sttement of the Zinga and not the assertion of te relation of Zinga
with te paka. This relation, kown as pakadharata, is conveyed only by
Te Buddhist Teor of Inference 97
llpanaa, and tus upanaa is not a duplication of hetu . This reply of Srldhara
doesn' t seem to be pertinent because it is comonly recognized tat het is the
statement of pakadhannatt.
Vacasapti and J ayanta flly realize the cogency of the argument as put
forward by Santarakita and the weakness of Uddyotakara' s defense. But
they once again appeal to psychological evidence. Their contention is that
our argument should be advanced in te same order in which we arrive at
truth i our own experience. Now, we frst observe the paka, say, "te hill"
and then the hetu " the smoke . " Next we remember the invari abl e
concomitance of smoke and fire as exemplified in the kitchen, and
immediately tis ivariable concomitance is understood i relation to smoke
on te hill , and te direct result of this is .te deduction of te conclusion.
Thus , the five avayavas enbody only the order of our own subj ective
ratiocination and as such constitute te most satisfing process of anumtna.
It must, however, be stated in favor of Buddhist contention that the
psychological account of the process of anllmtna, should not be confsed with
the logical requirement of the trut and validity of anllliana upon which alone its
evidential value depends. Psychological evidence as to the subjective process of
reasoning should not be allowed to dictate terms to te constituents of a logically
sound anlmtna. Logic, as has been pointed out, is concerned wit validity and
cogency, whereas psychology is interested in the natural history of mental
phenomena. So, though psychologically prtiit and llpanaya may have a
raison d' etre, tey cannot on tat account be tken to be avaavas.
A regards the nigamana, Buddhists insist that tis also doesn' t require to be
stated, as it follows by a sheer logical necessit from te universal proposition as
stated in the ldtharala and from the statement of hetu. Moreover, on te
Naiyayika' s, own showing, it is perfectly redundant, beig only a repetition of
p rtiit. 87
Uddyotakra, however, argues that nigamana is not a useless repetition
of prtijit, as te latter states only what is .yet to be proved, and the former
is the statement of the proven fact ; nor is the statement of nigamana
unecessary because it serves to dispel a likely error or doubt. Unless te nigal1lana
is stated, the opponent may still waver as to whether the prtiis acceptable or
not . The clear statement of prtijit dispels all such doubt and satisfes te opponent
completely. Santarakita points out tat this defense of the Naiyayikas is not
forcefl . In fact, there can be no possibilit of doubt if the hetu satisfes all the
three logical requirements, that is, if it is trirua linga. The conclusion
irresistibly follows from tis and from none else. If, on the oter hand, the three
rapas are lacking, the statement of nigamana by itself cannot remove doubt
about its validity. Santarakita refers to Aviddhakala, who contends that te
premises, beig scattered and piecemeal, cannot establish a unitary judgment
unless tey are shown to be mutally related, and tis relationship is shown by
te conclusion. To tis he replies tat tough the premises are scattered and
piecemeal, they are kit together by a logical bond of mutal compatibility and
98 Buddhist Eistemology
relevanc, and therefore the nigamana is unnecessary even for the purpose of
showing the mutual relation of the premises. 88
Because of the foregoing considerations Buddhist logicians reduce the
avayavas to two, namely, hetu and udlharaFa, and ultimately these two to one,
namely, hetu, depending on the grasping power of the opponent.
It is proper to conclude this account concerig the number of avayavas
with the words of S . Mookherjee. "From the doctrine of ten membered syllogism
reduced to fve in the Nylyasutr and still frther reduced to two (or one) in
Buddhist Logic, we can trace the history of the evolution of syllogism. Naturally
the psychological and logical factors were mixed together in the doctrine of ten
membered syllogism. In the Nayiyayika' s syllogism there has been a bold attempt
to shake of the psychological incumbus, but still the psychological infuence did
not cease to be at work. In the Buddhist syllogism as propounded by Dilnaga
and Dharmaklrti, the psychological factors were careflly eliminated and the
syllogism received a perfectly logical shape. But the survival of the example was
a relic of the ancient sway of psychology and this was destined to be
unceremoniously brushed aside by the onslaughts of Jaia logicians. " 89
Avayavadvay Tadition
Though the Buddhist accout of te nuber of avaavas can be traced back
to pre-Diiaga thinkers, 9 for want of some defnite iformation we have to
begin witll te Naa Bindu of DharmakIrti . In the Naa Bindu free avaavas
of anumlna are jointly referred to as stdhara. The tree avaavas are nanled
pak$a, hetu, and dr$tlnta. The acceptance of paka, tat " is, . prtift, indicates
tat DharmakIrti is not aversed to te avaavatraa tradition.
Here, tough we fnd the statement of prtijfl it is not, i fact, a part of te
process of anumlna.
91
Both in the PrmlFa Vlrika and Naa Bindu we fmd
elaborate argulents advanced by Dharmaklrti to poit out the nonacceptance of
prtifl as an avaava of anumlna.
9
2
The main argulent put fortll by him is
that tlle statement of prtifl may be made, but it can be done away wit also. In
tllOse cases where prtifl is not stated, it is uderstood as inlplicit. In fact,
from the context itself te reference to prtifl is to be uderstood, even tough
it is not explicitly stated. Thus, only het and udlharFa are parts of the process
of anumlna. In fact , DharmakIrti goes a step frter and states tat udlharmJa
is needed for ordinary minds, but for an intelligent person only het suffices.
93
In the statement of parlrthlnumlna DharmakIrti advocates te model of
avaavatraa, consisting of prtijfa, UdlharFa, and upanaa, but maintains
that only udlharFa and upanaa are real avaavas because tey alone have
probative value. The exanlples of te three types of anumana are:
1 . Svablvanwnana
Udaharfa
Upanaya
Ngamana
Everything that exists is momentary, like a jar.
The sound exists.
Therefore sou
n
d is momentary.
The Buddhist Theory ofInference
2. Ktrtnwlltna
Udtharla
Upanaa
Nigamana
3 . Anllpalabdhanumana
Udaharla
Upanaa
Nigamana
99
Wherever there is smoke, there is fire, like in
kitchen.
Here there is smoke.
Therefore here tere is fre.
Whatsoever is present and is perc
"
eptibl e is
necessarily perceived, like a jar.
But tere is no such j ar beig perceived.
Therefore a jar is not present here.
94
For Diiaga and Dhanaklrti the nigamana is the stdha for which hetu
and udtharala or udtharG1:za and upanaya are the stdhana. Therefore they are
referred to as stdhantlga (constituents of proof , whereas prtiit and nigamana
are kown as astdhantlga. In the case of svtrthtnumtna, hetu and udthar{la
are stdhantlga and prtiit, upanaya and nigamana are astdhanalga, whereas
i case of partrthtnumtna, prtiit, hetu, and nigamana are stdhantlgas, and
udthar{la and upanaya are astdhantlgas. Thus, the Buddhist position is to be
regarded as avayavadvayi .
In conclusion, for Buddhists a statement of anumtna is ordinarily
expressed either as prtiit-hetu-udthrla or as udtharla-upanaya-nigamana.
But fom a strict logical poit of view neither prtiit nor nigamana is a genuine
part of the process of anumtna, and therefore the real process of anumtna consists
of only two avayavas, namely, hetu and udtharla or udtharla and upanaya.
This is a view advocated by Dhanaklrti and accepted by all of his followers, like
Santarakita and Mokakaragupta.
Appendix 1
Nyiyapravesaksutram
Stdhanal dUalam caiva sabhtsar parasarvide
prtakamanumanam ca sabhtsaril tvatasamvide.
Iti ststrarthasamgrhaJz.
Tatra pakadivacanani sadhanam. Pakahetudrnantavacanai rhi
prasnikanamaprtlto 'rthal prtpadyata iti. Tatr paka( prsiddho dharf
prsiddhavisealJa visitataa svaam stdhatvenepsital. Prtaktdyavinddha
iti vakaseaJ. Tadyath. NtaJz sabdo 'nito veti.
Hetustnrupalz. Kili punastrairpam. Pakadhamzatvam sapake sattvam
vipake casattvami ti . KaJz punafl sapakalz . Ko va vipaka i ti .
Stdhadharasamanena samano 'rthal sapaksah. Tadath. Anite sabde sadh e
ghatadirnitaJ sapaaJz. Vpako yatr stdham nasti. Yannitam tadakrtakam
drtam yathtkasamiti. Tatra ktakatvalil praatanantarfakatvam vasaaa evasti
vpae nse. Itaitad hetl. Dtto dvivdhaJz. Stdhanela vaidhamrela
c. Tata stdhannela tavat. ftr hetolz sapaa etttvam khapate. Tadyatht.
Yatkrtakam tadanitam drtam yatht ghatadiriti. Vaidharmyelapi . Yatr
stdhabhtve hetorbhtva eva kathate. Tadyatht. Yannitam tadakrtakarl drt
ar yathtkasamit. Nitasabdenatranitatvaabhtva ucate. Akrtakasab
d
enapi
krakatvasabhtvafl. Yatht bhavabhtvo 'bhtva it. UtaJ paktdaafl.
Eam vacanani parpratayanakale sadhanam. Tadyatha. Anital
sabda iti pakavacanam. Krtakattditi p
i
kadhamzavacanam. Yatkrtakam
tadanitam drtam yatht ghatadiriti sapakanugamavacanam. Yannitam
tadakrtakam drtam yathtkasamiti vatirekavacanam. Etaneva trao 'vaava
itucante.
Sthaitumito 'i prtaadivirddhafl pakabhtsah. Tayatht. (1) Pata
aviruddhalz, (2) Anumanavirddhafl, (3)
A
gamaviruddhafl, (4) Lokaviruddhab,
(5)_Svavacanaviruddhab, (6) Apriddhaviseazab, (7) Aprsiddhaviseab, (8)
Aprasiddhobhaalz, (9) Pasiddhasambandhaceti. Tatra Patakavirddho yatht.
Asravalab}abda iti. Anumanavirddho yatht. Nito ghata iti. Agamavirddho
yatha. Vaiseikasya nital sabda iti stdhaatal. Lokavinddho yatht.
S
uci
1 02 Appendix 1
narsirfzkapilam prizyargatviccharkhasuktivaditi. Svavacanavirddho yathi.
Miti me vandhet. Apraiddhavisesano yathi Baudhaa Simkham prti vinasl
sabda iti. Aprasiddhavisero yathi. Stlizkhasa Bauddham prt cetana ateti.
Aprasiddhobhayo yatha. Vaiseikasya Bauddham prati
sukhidisamavaikirarJamieti. Prasiddhasambandho yathi. SriVa1a/ sabda
iti. Eim vacanani dhamwsvanlpaniriar1amukhena prtipidanaambhavatafl
sidhanavaiphalyatasceti prtiiidoa/. Uta/ palibhisi/.
Asiddhinaikintikaviruddhi hetvibhisih. Tatrisiddhascatu/prkir/.
Tadathi. (1) Ubhaisiddhal;, (2) Anatarisiddhafz, (3) Sariiddha/, (4)
Asraisiddhaieti. Tatr sabdanitatve sadhe cikuatvaditubhaisiddhah.
lrtakatviditisbdibhiva- ktividinaliz prtanatarisiddha/. Bipidibhivena
sandihamano bhUtamighito " gnisiddhivupadiiamina/ salldigdhisiddhah.
Drvamakisam gU1asraatvidita- ktsasattavidinaJi prtasraasiddha/.
Aaiantikal 1)atprakara. ( 1 ) SadharaJa, (2) AadharaJa, (3) Sapak1
aidesavrt irvipak1avyapI, (4) Vipa1aidesavrtsapa1avyapI, (5) Ubhayapak1)
-aikadesavrti, (6) ViruddhavyabhicarI ceti .
Tatra stdhirnah sabda/ prmeatvinnita iti. Taddhi nitanitapakaob
sadhara1atvadanaikintikam. Kim ghaJCvatprameyatvidanital; sabda
ihosvidikasavatrmeatvannita iti. Asidhiranah srava1atinnita iti. Tdi
ni tyinityapakibhyiliz vyavrttatvinni tini tyavini rmuktasya
cinyasyasambhavttsamsayahetub. Kimbhutasyisya sriva1atvamiti. Sapaks
aikadeiavrttirvipaksavyapl yathi. Apraatninantarlyakab sabdo 'nitatvat.
Apryatinantariyakal; pakafl. Asa vidudikasidib sapakal; . Tatrikadese
vidyudidau vidyate 'nitatam nikasidau. Apraatanantariyakal pakal. Asa
ghaJidiripalab. Tatra saratr ghatidau vidyate 'nitatvam. Tasmidetadapi
vdughaJaihamce1 naiki nt. Kim ga nitaitraatitla
sabdafl ihosvidvidyudidivadanitatvidapryatninantarlyaka iti. Vipaks
aikadeiavrttih sapaksavipl yathi. Praatinantarlyakafz sabdo 'nitatvit.
Paatinantarfakab pala/. Asa ghaJidi/ sapaka/. Tatr sarvatr ghaJidau
vidate " nitatam. Paatinantarrakal; pakal. Asa vidyudikisidlrvipak
a/. Tatrikadeie vidyudidau vidyate 'nitatvam nikasidau. Tasmidetadapi
vidyudghaJasidhanew purvavadanaikintikam. Ubhayapaksaikadeiavrttir
yathi. Nitalz sabdo 'murtatvidit. Nitab palab. Asikia parmi 1vidi/ sapal
all. Tatrikadeia ikisidau vidyate 'murtatvam na parmiwu. Nital pakab.
Asa ghaJa sukhidirvipalafl. Tatraikadeie sukhidau vidate " murtatvam na
ghaJidau. Tasmidetadapi sukhikiSasidharmye1inaikintikam.
Virddhivabhicarl yathi. Anitafz sabdal; krakatvid ghaJavat. Nitalz sabdafz
sriVa1atvit sabdttvavaditi . Ubhayo{ sa7isayahetutvad
dvivapetiveko 'naikintika/ samuditiveva.
Virddhaiatu/prkirl. Tadathi. ( Dharasvanlpaviparitasidhanafl,
(2) Dhamwviseaviparltasidhanal;, (3) Dharisvar-paviparitasadhana/, (4)
Dhamzivise-avipartaidhanacet. Tatra dhamwsvar-pavipantastdhano yathi.
Nital sabda/ krtakatvtt praatinantarlyakatvidveti. Ami heturvipaka eva
bhividvirddha/. Dhamwvisesaviparitasidhano yathi. Pararhicakuridaab
Appendix i 103
sarzghttatvtcchaantsantd- yangvise$avaditi: Aarz heturatht ptrtnham cak$
urtd/ntm sthayati tatht sarz hatatvampi parasyttmanal) stdhayati .
Ubhaattvabhictrtt. Dhamzisvanpaviparftastdhano yatht. Na dravarz na
kanna na glllO bhtvab ekadrvavattvtt glakamzasu ca bhtvtt stmtnavise$
avaditi. Aymh hi heturatht drVldiprti$edhmi bhtvasa stdhaati tatht
bhlvasytbhlvatvamapi sldhayati. Ubhayatrlvyabhictrtt. Dharmivises
aviparftasldhano yatha. Aameva hetrsminneva purva-pak$e 'saiva dhamzi1O
yo vise$al) satrtaakarm nlma tadviparftamasatrataa- kanrvamapi
slaati. Ubhaattvabhicartt.
Drstlntlbhlso dvividhal). StdhamelJa vtidhareza ca. Tatra stdhannelJa
tivad dT!antabhtsal paicaprakarl. Tadyatht. () Sadhanadhantsiddha/z,
(2) Sadhadhanasiddhal), (3) Ubhaadhanntsiddhal), (4) Ananvaa/z, (5)
Viparftanvaascet. Tatr stdhanadhamzisiddho yatha. Nita/z sabdo 'munattt
parmalJuvat. Yadamunmi tannitam dr$!arz yatha paramtlJul). Parmtzau hi
sadhymiz ni tyatvamasti stdhanadharmo 'murtatvaliz nasti
munatvttarmazuntmiti. Stdhadhanisiddho yatha. Nita/z sabdo 'mrmatvtd
buddhivat. Yadamurtarz tannitaJi dr$!arz yatht buddhilz . Buddhall hi
stdhanadharmo 'munatvamasti sadhadharmo nitatvarz nasti. Anityatvtd
blddheriti. Ubhaytsiddho dvividhal) . Sannasanketi. Tatr ghatavaditi
vidyamtnobhaytsiddha! . Ani tyatvtnmurtatvtcca gha!asya.
Aklavaditavidyamanobhatsiddhal). Tattvavtdinarz prti. Ananvao yatra
vinanvayena stdhyastdhanayol) sahabhaval) prdarsyate. Yatha gha!e
krtakatvamanitatvam ca dr$!amiti. Viparfttnvao yatha. Yat k.nakmiz tadanitmiz
dr$!amiti vaktave yadanitam tatkrtakam dr$!amiti brvfti. VaidhamrelJapi dr$
!anttbhlsaz paicaprktrl) Taatht. (1) Sldhavav.rt ab, (2) Stdhantavrtal),
(3) Ubhaytvtvrttal, (4) Avyatirekal), (5) Viparftavatirekaketi. Tatr
stdhyavyavrtto yatht. Nita/ sabdo " murtatvtt parmalJuvat. YadanitaJh
tanmzrtmh dr$!mh yatht parml1l/. ParmalJorhi stdhanadhanno " munatv
vtvrttam minatvatarmalJunamiti. StdhadhamlO nitatvmh na vivrtarz
nitatvatarmilJinamiti. Stdhanavtvnto yatht. Kamzavaditi. Kannalal
siarz nitatvam viv.rtam. AnitatvitkanalJal). Stdhanadhanno 'munatvam
na vyavrttam. Amirtatvatkarmala/. Ubhaytvyavrttah. Akasavaditi .
Tatsattvavadinam prati . Tato nitatvamamurtatvam ca na vytvrttam.
Nitatvtdamunatvaccilaseti. Avatireko yatr vint stdhastdhananivrtl
tadvipak$abhivo nidarfate. Yathi gha!e murtatvamanitatvmiz ca dr$!amiti.
Viparftavatireko yatht. fitmi tanmunmiz dr$!amiti vaktave yanmunmh
tadanitmi dr$!amiti brvfti.
E$am pak$ahetudr$!antlbhtsanim vacanini sldhantbhtsam.
Atzraai rhaz t prm olna ca dY e pli . Tatra prtak
a kalpanapodham yajjiinamanhe nptdal namaattdikalpanarahitam. Tada
amak$arz prti vanata iti prtak$am. Anumanam lingtdanhadaranam. Lingmi
plnastrinipamuktam. Tasmldadanumee 'nhe jitnamltadyate 'gnirtra anita/z
sabda iti vi tadanlminam. Ubhaatr tadeva jiinam phalamadhigamanpatvtt.
Savyaparvatkhyttelz prmilJatvamiti. Kalpantjiinamarthintare prtaks
104 Appendix 1
tbhisam. Yjjitnaln gha!al) pa
.
t iti va viklpaatal) samutadate tadarthavala
Wzavi$,aatvttrtak$ibhtsam. Hetvibhapzrvakrzjitnarz,anumtntbhisam.
Hetvibhiso hi bahuprktr uktab. Tadanumee 'rthe jitnanzavutannasa
bhavati tadanumantbhisam.
Sidhanado$odbhtvanani dlt. Sdd$o na. Pa$a pmt
, idiviruddhatvam. Hetudo$o 'siddhinaikintikaviruddhatvam. D!$!tntado$al)
sidhana,dhanntdyasiddhatvam. Tasyodbhivanam prtnikaprtiyanmi dl$
wzam. Abhltasidhanado$odbhivantni dlsanibhtsini . Samplr{ze stdhane
ny lnatvavacanam. Adu$!apak$e pak$ado$avacanam. Siddhahetuke '
siddhahetukariz vacanam. Ekintahetll ke ' nekantahetlkarz vacanam.
Avirlddhahetlke virlddhahetukmiz vacanam. Adu$!ad!$!tnte dU$!ad!$!
antado$avacanam. Ettni dU$a{ztbhistni . Na hyebhil) parapak$o dU$ate.
Nirvadyatvattasa. Ituparmate.
Padtrhamatrmakhatamidal Dilmatrsiddhae;
Yttr yuktiralktirva sanatr slvicarita.
** Iti N
A
YPRV
S
AKS
U
M samiptam**
Appendix 2
Nyayapravesaksutram
(Traslation)
Giving arguments in support of one' s own position and pointing out defects in
the rival ' s position along with their respective fallacies are essential for
commuicating kowledge to others, whereas for acquiring kowledge for one' s
own sake, perception and inference and teir respective fallacies are essential .
This is te sum and substance of the science of debate. 1
Here, te statement regarding subject (aka) and so on is means (sadhana) . 2
The statements of subject, reason (het) , and example (drtanta) establish te
hitherto not kown object to the inquirer. Subject is a kown substratu having
the kown qualifing property through which another propert is desired to be
proved by someone.
3
The subject must not be contrary to perception and so on.
In the example of whether sound is eteral or noneteral , the subject is sound,
and the propert tat is desired to be proved is wheter it is eteral or noneteral .
This is te point of debate.
Reason has tree featres.
4
Now the question is, Which are these tree
featres? The necessary presence of reason in te subject, te presence of reason
only in homologue (sapaka) , and te necessary absence of reason always in
heterologue (vipaa) are te three features of reason. Again, what is homologue?
5
What is heterologue? Homologue is what is similar (to subject) by virte of
having probandum (sadha) as its common property. For example, in order to
prove tat "Sound is noneteral , " pot and so
.
on, which are noneteral , become
homologue. Heterologue is that where probandum is absent for example, in the
inference "Watever is eteral is experienced to be nonproduced as space and so
on. " Being produced or being necessarily asspciated wit efort is present only
in homologue and is never present i heterologue. Here, reason is noneteral and
so on.
Example is twofold.
6
One is based on similarity (sadhane1a) and, the
other is based on dissimilarity (vaidhane1G) . That which is based on similarity
106 Appendix 2
is where the existence of reason is kown only in homologue. Its example is.
"Watever is produced, is experienced as noneteral , for example pot and so
on. " That which is based on dissimilarit is where absence of reason is said to be
in the absence of probandum, for example: "Whatever is eteral is experienced
to be nonproduced as space and so on. " "Eteral" means absence of non eteral .
The word "nonproduced" means absence of being produced, as for example,
"nonbeing" is absence of being. Thus, we have stted subject and so on.
Statements about all these are means of communicating kowledge to oters.
That can be explained as follows : "Sound is noneteral " is a statement of subject,
and "because it is produced" is a statement of subject having reason as its propert
(pakadharmata) . "Watever is produced is experienced to be noneteral like
pot and so on" is the statement of homologue, which is based on similarity. The
statement of dissimilarity (vatireka) is "Whatever is eteral is experienced to be
nonproduced like space ad so on. " These are said to be te teeleent (amas)
of te prs of amentaton i sppor of one's o psiton.
Even though it is desired to be proved, a fallacious thesis (akabhasa) is
(fallacious because it is) contrary to perception and so on. 7 11 is of the following
nine types : ( 1) contrary to perception (prtaka viruddha) , (2) contrary to
inference (anumana viruddha) , (3) contrary to one' s own scriptre (agama.
viruddha) , (4) contrary to public opinion (toka virddha) , (5) contrary to one' s
own statement (svavacana viruddha) , (6) a thesis whose subject is unacceptable
(aprsiddha viseWlG) , (7) a thesis whose predicate is unacceptable (aprasiddha
viseya) , (8) unacceptable to both parties (aprsiddhobhaya) , and (9) a thesis too
well kown or accepted universally (rasiddhi sambandha) .
(i)
A thesis contrary
to perception is like " Sound is audible.
"(ii)
A thesis contrary to inference is like
"Pot is eteral .
" (iii)
A
thesis contrary to one' s own scriptre is like a Vaiseika
adherent trying to prove that "Sound is eteral . "
(iv)
A thesis contrary to popular
belief is like "A dead man' s skull is pure because it is a limb of an animate being,
as, for example, a conch or a shell .
"(v)
A
thesis contrary to one' s own statement
is like " My mother is barren.
"(vi)
A tesis whose predicate is unacceptable is like
while arguing with a Sarya adherent, a Buddhist takig his stand on such a
proposition as " Sound is perishable.
" (vii)
A tesis whose subject is unacceptable
is l, while arguing with a Buddhist, a Sarya adherent taking his stand on
such a proposition as "Self or ego is animate.
" (viii)
A tesis whose both the terms
- the subject and the predicate - are unacceptable is like a Vaiseika adherent
arguing with a Buddhist and taking his stand on such a proposition as "Soul is the
inherent cause of pleasure and so on.
"(ix)
A
thesis universally accepted is like
"Sound is audible. "
These statements have a fallacious thesis (rtiftdoa pa!tbhtsa) because
they canot be logically established and, second, they canot serve as a means of
proof. They negate the very property that is to be proved. Thus, fallacies of
thesis are stated.
Unproved (asiddha) , inconclusive (anekantika) , and contradictory (virddha)
are the fallacies of reason (hetvabhtsas) .
8
Appendix 2 107
Unproved is divided into four types : ( 1) unacceptable to both (ubhatsidha) ,
(2) unacceptable to eiter (anyatarasiddha) , (3) unproved because of doubt
(sarigdhtsiddh'a) , and (4) unproved locus (try
a
siddha) .
In the argument " SOlnd is noneteral on account of being visible, " sound is
subject, noneteral is probandU1, and visible is reason. Here visibility (of sound)
is unacceptable to bot parties, and, terefore, it is a fallac of being unaccepable
to both. Again, the inference " Sound is noneternal because it is effort-made, "
will not be acceptable to a person upholding the position that sound is eternal a
that only its manifestations, which are effort-made, are noneternal . On the basis
of doubtfl existence (of reason) , if one undertakes the inference like, inferring
fre on te basis of doubting cluster of mist as smoke, ten it is a fallac of
doubtfl existence. "Space is a substance because it is a locus of quality (sound) " :
(this argument is valid) for believers in the reality of space; but (for Buddhists) it
is not valid because (Buddhists) don' t believe in te realit of space.
Inconclusive is sixfold : ( 1 ) too wide (sadMra) , (2) too narrow
(asidhara) , (3) pervading one part of the homologue and pervading the entire
heterologue (sapak$aikadefavril vipak$avapl,
9
(4) pervading one part of the
heterologue and pervading the entire homologue (vipak$aikadesavrttil
sapak$avapl ,1 (5) pervading one part of both (ubhayapakaidesavrtti) 1 and
(6) contrary and nondivergent, (virdhavabhicarl.
1
2
The exmple of "too wide"
is that " Sound is eternal on account of its knowability. " Here, the reason
(knowability) is common to bot eteral and noneteral tings ; therefore it is a
fallac of inconclusiveness of te variet of being " too wide. " Is sound noneteral
like a pot on account of its knowability or is it eternal like space on account of its
knowabilit? The example of " too narrow" is "Sound is eternal on account of its
audibility. " Here, audibility is peculiar to sound and is not found in any other
thing eteral or noneternal , and there is no other thing different from eteral or
noneternal, and this gives rise to doubt as to what help audibility (as a reason)
can render. "Pervading one part of the homologue and pervading the entire
heterologue" is as follows : "Sound is natrally existing" (which can exist without
any volitional effort on our part) . Here, the terl "natural" (apryatanatarfyaka)
is subject. Its homologues are lightnin, space, and so on. Here in one part of the
homologue like lightning, noneternality resides, but it doesn' t reside in some
other parts like space and so on. Let us agai take the tenl "natural" as subject.
Pot and so on are its heterologues. Now this noneterality resides in all heterologues
like pot and so on. Therefore, being common to both lightning (which is
nonproduced) and pot (which is produced) , it is inconclusive. Here a doubt
arises whether sound is efort-made because of noneternality like pot or noneffort
made because ofnoneteralit like lightg. "Residing i some part ofheterologue
and residing in whole of homologue" is as follows : "Sound is effort-made because
it is noneternal . " Here subject is effort-made. Pot and so on are its homologues.
Everywhere in pot and so on noneternality resides. Again subject is effort-made;
its heterologues are lightng, space, a so on. Here, i one part of the heterologue
like lightning, noneternality resides but does not reside in other part like space
108 Appendix 2
and so on. Therefore, tis also being coran to bot lightnig and pot it is
inconclusive as before. "Pervading one part of both (i . e. , homologue and
heterologue) " is lie "Sound is eteral because it is incorporeal . " Here subject is
eterality. . Space, infinitesimal particle, and so on are its homologues. In one
part of te homologue, tat is in space, incorporealit resides but not in the other
par, that is, an infitesimal particle. Agai, subject is eteralit. Pot, pleasure
and so on are its heterologue. Here, in one par of the heterologue, that is,
pleasure incororeality resides but not in the oter part of the heterologue like
pot and so on. Since it is coran to bot pleasure (which is heterologue) and
space (which is homologue) , it is inconclusive. A example of " Contrary and
nondivergent " is as follows : "Sound is noneternal on account of being produced
like a pot" and "Sound is eteral on account of audibility like soundness, " i
bot tese arguents, te two reasons taken togeter cause doubt and are terefore
inconclusive (here, about the same tesis two diferent reasons lead to two
contradictory conclusions) .
Contradictory is of four tpes. They are as follows : ( 1 ) reason contradictory
to probandum (dhannasvarpaviparlastdhana) ,
13
(2) reason contradictory to a
paricular probandum (dhanaviseaviparftastdhana) ,
14
(3) reason contradictory
to thesis (dhamvanavrparitastdhana) ,
15
(4) reason contradictory to a paricular
tesis (dhanniviseaviparrtastdhana) .
1 6
"Reason contradictory to probandum"
is as follows : "Sound is eteral on account of being produced or efort made. "
This reason belongs to the side of heterologues. Therefore, it is lown as
contradictory. "Reason contradictory to a particular probandum" is as follows :
eyes and so on exist for the sake of others on account of their composite character
lie bed, seat etc. which are composite in nature. This reason not only proves
that eyes and so on exist for te sake of others but also proves that self (as
conceived by the Sarkhyas) is of a composite nature because reason is
invariably applicable to bot. "Reason contradictory to the tesis " is as follows :
"Summum genus (bhava) is not a substance or action or quality, because of its
possessing one substance and because of its presence in qualities and actions like
the universal and te particular. " This reason not only proves that summum genus
is the opposite of substance and so on but also proves the absence of summum
genus itself because it is invariably coran to both. "Reason contradictory to a
particular thesis " : tis very reason in tis very argument of te opponent proves
the presence as well as the absence of "being based on being" and "beig based
on nonbeing" and is coran to bot.
17
The fallac of examples is twofold, on te basis of siilarit and on the basis
of dissiilarit.
The fallacy of example based on siilarity is fivefold. It is as follows :
( 1 ) incompatibility of the example with te reason (Stdhanadhanntsiddha) ,
1 8
(2) incompatibilit of the example wit te probandum (Stdhadhanntsiddha) ,
1 9
( 3) incompatibility of the example with both (reason and probandum)
( Ubhayadharmasiddha) ,
2
0
(4) absence of statement of positive invariable
concomitance (Ananv
a
a) ,
2
1
and (5) contrary to the statement of positive invriable .
Appendix 2 109
concomitance ( Viparfanvaya/) .
22
"Incompatibility of example with the reason"
is as follows : "Sound is eteral On account of its being incorporeal lie infnitesimal
particle. " In infinitesial particle the probandum, that is, eterality, exists, but
the reason, namely, incorporeality, doesn' t exist because infinitesimal particles
are corporeal . "Incompatibility of the exanlple wit te probandum" is as follows :
" Sound is eternal on account of its being incorporeal Like intelligelce. " In
intelligence, the reason, that is incorporeality, exists, and the. probandum, tat is
eterality, doesn' t exist because intelligence is noneteral . Incompatibility of the
example with both (reason and probandum) is twofold, positive and negative.
The positive is as follows : "Sound is eternal on account of being incorporeal like
a pot. " Here, te example is contrar to both reason and probandum because pot
is noneternal (contrary to probandum) and corporeal (contrary to reason) . The
exmple of negative is as follows: "Sound is eteral on account of being incorporeal
like space. " This argument contains an example that is not acceptable to a person
who doesn' t believe in te reality of space. "Absence of statement of positive
ivariable concomitance" is as follows : the coexistence of probandum and reason
is stated witout stating teir invariable concomitance, for exmple in the statement
" In the pot there is te propert of being produced and noneterality. " " Contrary
to the statement of positive invariable concomitance" is as follows : instead of
saying, "Whatever is produced is experienced to be noneteral , " we say, "Whatever
is noneteral is experienced to be produced. " (Here concomitance is stated in a
reverse order.)
Fallac of example based on dissimilarity23 is fvefold, which is as follows :
( 1 ) nonabs ent probandum (Sadhyavyavrtta) ,
24
(2) nonabsent reason
(Stdhanavtv.rta) ,
25
(3) both are nonabsent ( Uhayavavrtta) ,
26
(4) absence of
statement of negative concomitance (Avatireka) , 27 and (5) contrary to the statement
of negative invariabl e concomitance ( Viparltavyatireka) .
28
" Nonabsent
probandum" is as follows : "Sound is eternal on account of incorporeality like
infmitesimal particle. Whatever is noneteral is experienced to be corporeal l
infnitesimal particle. " Of infnitesimal particle, reason is incorporeality, but
incorporealit, in fact, is absent fom infinitesimal particles because they are
corporeal, whereas the probandum, namely, eterality, is not excluded because
infnitesimal particles are eternal . "Nonabsent reason" is as follows : " Sound is
eteral on account of incorporeality like action. " Here, the probandum, that is,
eterality, is excluded i respect of action because all actions are noneteral, but
the reason, that is incorporealit is not excluded because an action is icorporeal .
"Both are nonabsent" is as follows: "Sound is eternal on account of incorporealit
like space. " This is fallacious for one who believes in the existence of space.
Here, both eternality (probandum) and incorporeality (reason) are not excluded
because space is both eteral and noncororeal . "Absence of statement of negative
concomitance" is that where without negating probandunl and reason, their
presence in te heterologue is stated, as, for exmple, in pot there are incorporealit
and noneterality. "Contrary to the statement of negative invariable concomitance"
is as follows : instead of saying "Whatever is noneteral is experienced to be
1 10 Appendix 2
corporeal, " we say, "Whatever is corporeal is experienced to be noneteral . "
These are the statements of the fallacies of thesis, reason, and example, and
they are known as fallacies of demonstration or means of proof.
In order to acquire kowledge for one' s own sake perception and inference
are the only two means of kowledge. Perception is beref of conceptalization.
It is that kowledge that is devoid of all deterinations in terms of words and
concepts like class and so on That which depends on each individual cognitive
sense is perception. Inference is knowledge of objects on the basis of a necessary
mark. Necessary mark is said to have three features. Therefore, it is that kowledge
that arises concering the inferential object, for exmple, " Here is fre (because
of smoke) " or "Sound is nonetemal (because it is produced) . " In both (perception
and inference) the same kowledge is both the cause and its result, which is in
the form of cognition of an object. Knowledge is always of fnctional character,
and, therefore, it is true cognition. (The truth of kowledge consists in its ability
to be fnctional . ) A determinate kowledge is a fallacious perception if it is
taken in the context of objective reality.
The object of determinate kowledge is a universal, but if a unique particular
i taken to be its object, then it is fallacious. So, (determinate) knowledge like
"This is a pot" or "This is cloth, " which arises through a judgment, is fallacious
perception because it takes the objective reality (unique particular) as its object.
Fallacious inference is that which is due to fallacious reason. Fallacious reason is
said to be of many kinds. Thus, concering the inferential object, that knowledge
that arises in one who doesn' t take cognizance of (fallacies like unproved and so
on) is fallacious inference.
To point out defects in means of proof is known as fallacies in means of
proof or reftation (daala) . Defciency in means of proofs is (known as) its
defect. Defect in thesis consists in incompatibility with perception and so
on. Defect i reason is unproved, inconclusive, and contradictory. Defect in
exmple is where reason and so on as property are not proved. To point out
the defects to the inquirer is for his or her knowledge.
To point out that the defects are not, in fact, defects is kown as fallacious
reftation. To point out deficienc in a perfect reason, to point out defects in
nondefective thesis , to point out unprovenness of a proved reason, to point
out inconclusiveness in a conclusive reason, to point out contradiction in a
noncontradictory reason, to point out defects in a nondefective example-these
are fallacious reftations. By these the rival position is not vitiated because they
are not reftable. This is the end.
The entire discourse on (logical) categories is stated to prove the philosophical
position of DiInaga. Whatever is argued or not argued here is well discussed (in
Prmala Samuccaya) (by us) .
Notes
CHTR 1
1 . In Buddhist epistemology the Sanskrit word for knowledge is prmo1a, and the
words prmo, samyag-jitna,. ad so on ae also used as its synonym.
2. Dharmaklrti , Naa Bindu, p. l .
3 . Dhaaklrti , Pamt1a Vtrtik, p . 3 .
4. Ibid. , p. 8 .
5 . Dharottara, Naa Bindu Tm, p. 10.
6. Diimaga, Prmt1a Samllccaa, Tr. by Hattori, Dilntga on Perception, p. 28.
7. Ibid, p. 29.
8. Dharmottara, Noya Bindu Tm, pp. 82-85.
9. Aniruddha, AbhidhammamatsaJngaho, IV. 8.
10. Edited by Kashyap, Jagdish, Kathvatthu, p. 374.
11 . T. Stcherbatsky, Buddhist Lgic, vol . 2, p. 311 .
1 2. Ibid. , Appendix-IV.
1 3. Ibid. , vol . 1 , p. 552.
14. (a) Diimaga, Pamt1a Samllccaa, Tr. by Hattori, Dilntga on Perception,
p. 28.
(b) Dharmaklrti, Pramt1a Vtrtika, pp. 479, 480, 491 .
1 5. Dharmaklrti , PaJnt1a Vtrtika, p. 337.
1 6. Diimaga, Prmtza Samllccaya, Tr. by Hattori, Dilntga on Perception, p. 28.
1 7. Ibid. , p. 29.
1 8. Ibid. , p. 108.
1 9. Ibid. , p. 28 .
20. Ibid. , p. 29.
21 . Ibid. , p. 28 .
22. Dharmottara, Ntya Bindll Tm, pp. 69f.
23 . Dinnaga, Pamt1a Samuccaa, Tr. by Hattori , Dilntga on Perception, p. 30.
CHTR 2
1 . Diimaga, Prmt1a Samllccaya, Tr. by Hattori, Dilntga on Perception, p. 32.
2. See Bhatt, Basic Ways ofKnowing, p. 31 .
3 . Diimaga, Prmt1a Samllccaa, Tr. by Hattori , Dilntga on Perception, p. 24.
4. Ibid, p. 25.
1 12
Notes
5 . Tatr sand/ne na prma1lantarm anietasake/, ibid. , p. 24.
6. Ibid. , p. 28.
7. Ibid. , p. 29.
8. Ibid. , pp. 25-26.
9. Vasubandhu, Abhidharmakofa, p:145 .
10. Dhanalrti , Nia Billdu, p. 32.
11 . Dharmottara, Nia Bildll Tka, pp. 12ff.
1 2. Dhanaklrti, Naya Bildu, p . 42.
1 3 . Ibid. , pp. 42f.
1 4. Cf. avapadefam of the Nyaya school .
1 5. Dharmottara, Naya Bindll Tka, p. 44.
1 6. Viklpa yonaa/ fabdtfca fabda vikalpa yonaa/.
1 7. Dharmottar, Nia Billdu TM, pp. 32f.
1 8 . Ibid. , pp. 51f.
1 9. Ibid. , pp. 54.
20.
S
atarakita, Tatta Smngrha, verses 1220-23.
21 . Ibid. , verse 1 223
22. Ibid. , verse 1224.
23 . Ibid . , verse 1236.
24. Ibid. , verse 1 31 2.
25 . Th. Stcherbatsky, Buddhist Logic, vol . 1 , pp. 1 53-45.
26. Dhanottara, Nia Bindll T, p. 36.
27.
S
antarakita, Tattva Smngrha, verse. 1 31 3.
28. Dhanottara, Naa Bildu TM, pp. 36-40.
29.
S
antarakita, Tattva Smngrha, verse. 1 323 .
30. Kamalasila, Tatta Smizgrha Panjika, p. 482.
31 . Dharmaraja Adhvarindra, Vedanta Paribhta, p. 32.
32. Bhartrhari , Vakapad;ya, I, verses 123-24.
33. ISvara Krla, Stzka Karika, pp. 27-28.
34. Bhatta, Kumarila,
S
loka Vartik, prta!a khmida, p. 1 33.
35. Ibid. , p. 73 .
36. Bhatta, Jayanta, Naya Manjar;, p. 92.
37. Visvanatha, Bhta Pariccheda, p. 58.
38. Diiaga, Pama1a Samllccaa, Tr. by Hattori , Dilinaga Oi Perception, p . 88.
39. Bhartrhari, Vakapadfya Ch. II , verse 48. Kamalasila i n Pafjika quotes this
verse in order to substantiate the Buddhist position, which
S
intarakita presents in Tattva
Smizgrha in another verse with similar import.
40.
S
atarakita, Tattva Smigrha, verses 1292ff.
41 .
S
antarakita, Tattva Smizgrha, verses 1268-90.
42 . Diimaga, Prama1a Samuccaya, Tr. by Hattori , Dilinaga Oi Perception,
Appendix 1 . 5 . Atr vifea(zam parmatape!am sarvetvavikalpaM ea.
43 . The Buddhist theory of malasa prta!a is diferent fom tat of the Nyaya. In
Nyaya malasa prta!a is internal perception of the self and mental states through the
mind, which is an inner cognitive sense organ.
44. Diimaga, Prma1a Samllccaya, Tr. by Hattori , Dilllaga on Perception,
Appendix, I . 6.
45. Ibid.
46. Dharmottara, Naya Bindll Tka, pp. 46-48.
Notes 1 1 3
47. KamalaSla, Tattva Samgrha Paijik, pp. 1260-6l .
48. Dhanottara, Naa Bindu Tiki. pp. 419-50.
49. Ibid. , pp. 56f.
50. DhannakirtI, PilJa Vtrtik, p. 243.
51 . Ibid. , p. 239.
52. Dhanottara, Naa Bindu Tki, pp. 56f.
53. Ibid. , p. 60.
54. T. Stcherbatsky, Buddhist Logic, vol . 1 , p. 1 62.
55. Dharmottara, Ntya Bindu Tk; See Buddhist Logic, vol . 2, Appendix 3 .
56. Dhanottara, Naa Bindu Tki, p. 59.
57. Ibid.
58. Misra, Durveka, Dharmotar Padlpa, p. 62.
59. Dharmottara, Ntya Bindu Tk, pp. 62-63 .
60. Ibid.
61 . Dharmaklrti, PrmtlJa Vtrtika, pp. 268-69.
62. Ibid. , pp. 281-86.
CHTER 3
1 . Encclopaedia ofReligion and Ethics, vol . 8, p. 1 33, fn. 4.
2. Edited by Swai Dwarikadas Shastri, Sutta Nka, vol . II, p. 58, Digga Nka,
vol . III, p. 226.
3. Samuta Niki ya, vol . 2, p. 57.
4. Ibid. , p. 58.
5 . Asalga, Mahatnasutrtlmkr, 1. 12. ,
6. Sugiura, Sadajvio, On Some Aspects ofthe Doctnes of Maitreanttha and ASaJga,
pp. 52-53 .
7 . Ibid. , p. 74.
8 . Ibid. , p. 81 .
9. Vidyabhushan, Histor ofIndian Logic, p. 267.
10. Ibid. , p. 269.
11 . Th. Stcherbatsky, Soul Teor of the Buddhists, quoted from Buddhist Logic,
vol . 1 , p. 3l .
12. Uddyotakara, Ntya Vtrtika, p. 300.
1 3 . Diimiga, Prmila Samuccaa, Tr. by Hattori, Difmtga on Perception, IL L
14. Keith, Indian Logic and Atomism, p. 43 .
1 5. Chatterjee, Niya Teor ofKnowledge, pp. 244-45 .
1 6. Diimiga, Prm{l Samllccaya, Tr. by Hattori, Difmiga on Perception, I. 7-8.
1 7. Vicaspati , Ntya Virtik Titara Tki, p. 9.
1 8. Parisuddhi , Asiatic Society edition, p. m.
1 9. Dinniga, Prmtla Samllccaa Tr. by Hattori, Difmtga on Perception, I . 9.
20. Ibid. , I. 10.
21 . DhanakIrti , Ntya Bindll, 1. 1 .
22. Dharmottara, Ntya Bindu Tk, p . 131 .
23 . From te epistemic angle the relation can be named as gamak-gama-bhtva,
but from the ontological angle it can be put as gama-gamaka-bMva. In the text it is put
in te latter fon.
24. Dharmottara, Naa Bindu Tk, pp. lllf.
25 . Ibid. , p. 97.
1 1 4 Notes
26. DiInaga, Pamala Samllccaa Tr. by Hattori , Dilnaga on Perception, II. l .
27. Dharmaklrti, Nta Bindu, p . 98.
28. Dharottara, Naya Bindll Tki, pp. 98-99.
29. Th. Stcherbatsky, Buddhist Logic, vol . 1 , p. 236.
30. Dhanottara, Naya Billdu Tka, p. 97 . Pararthan1llllallam iabdCtmakam,
ivarthammzalam tZl jialttmakam.
31 . Iid . , p . 30.
32. Quoted from Rndle, Frgments fom Difntga, section 1 4.
33. Dhanakirti, Nta Bindu, p . 1 86.
34. Ibid.
35. Dharmottara, Ntya Bindll Tka, p. 97.
36. Dharmaklrti, Nta Bindll, p. 98 .
37. The word "individual " should not be understood as referring to svalaka(za.
Here it means every individual object belonging to the category of vikalpa or stmana
laala.
38. Dharmaklrti, Naya Bindll, p. 11l .
39. Vacaspati , Naya Vartika Tatpara Tka, p . 199.
40. Dharmaklrti, Naa Bindll, Ch. 2. pp. 122ff.
4l . Ibid. , p. 126.
42. Ibid. , pp. 117, 140.
43 . Prmala Vartika, p . 260 (Vrtti by Maoratanandi) .
44. Dharmaklrti, Naya Bindll , pp. 152ff.
45. Ballddha Tark Bhia, Mokaaragupta p. 62.
46. Randle, Frgments fom Difnaga, p. 46 .
. 47. Dhanaklrti , Nta Bindu, p . 129. Svabhiva prtibandhe hi satartho 'rtham
gamaet.
48 . Ibid . , p . 1 3l . Tadaprtibaddhasya tada vabhicarniyamabhivCt.
49. Dharmaklrti, Nta Bindu Tk, p. 129.
50. Hew Bindll, Dhanaklrti , p . 53 . Tasa vaptirhi vapakasa tafa bMva eva.
vapa va tattriva bhava. Etena anvao vatireko va.
51 . Arcata, Hetu Bindll Tk, pp. 1 7- 1 8.
52. Dharmaklrti, Naya Bindl l, p. 134.
53 . Dhanaklrti , Nta Bindu, pp. 134f.
54. Shah, Akalafka 's Criticism ojDharmakfrti 's Philosoph, p. 264, fn. 49.
55. Dhanaklrti Naya Bindll, p. 112.
56. Dharmottara, Nta Bindu Tka, Samanasa ca sa iabdadeaJ, p. 112.
57. Ibi d. , p . 11 3 . Na ca viieaJ sadhyaJ apitll samanam.
58. DharmakTrti , Nta Bindu, p. 11 4.
59. Ibid.
60. Ibid.
61 . T. Stcherbatsky, Buddhist Logic, vol . 1 , p . 243.
62. Uddyotakara, Nta Vartika, 1. 1 . 5 , p. 30l .
63. T. Stcherbatsky, Buddhist Logic, vol . 1 , p. 244.
64. Dharmottara, Nta Bindu Tk, p. 103 . Yadyapi cCtr niicitagrialam na k.ram
tatMpi ante kram prkantayordvaorpi rapaorpealiyam.
65. Dharmottara, Nta Bindll Tka, pp. 109- 10.
66. Buddhist logicias do not admit the possibility of kealanvai ad kevalavatirek
vtptis; the metaphysics of Naiyayikas, however, permit them to do so. In a simation
Notes 1 15
where tere is only te possibility of co-occurrence of heW and sadha in their presence,
and there is no possibility of their co-occurrence in absence, it is knon as kevalanvayi
vapti . Likewise, if in a situation tere is a possibility of co-occurrence of sadlra and hetu
only in their absence, and there is no possibility of their co-occurrence in their presence,
it is kown as kvalavatrek. The kevalanvai is purely aative concomitance, whereas
the kevalavatireki is purely negative concomitance. The exmple of former is, "Whatever
is kowable is nameable. " The example of latter is, "In ay pure substance different fom
earth, there is absence of earth because there is absence of smell . "
67. Establishing a contradictory thesis based on a rival reason.
68. Cf. Hemchandra, Pamt(la Mlst, p. 72, and Akaika, LaghfyastraaIII . 12,
published i n Sindhi Grata Mala.
69. Vadi Deva Suri , Pmt/JanaatattvaloktlaJikar, Ch. 3, p. 142.
70. Quoted by
S
antarakrita i n Tattva Smilgrha, verse 1 368 .
71 . For details see
S
antarakrita, Tattva Sanigrha, verses 1 363- 1428 .
72. Hew caka is also called palad!arma caka; see Vidyabhushaa, A Hstor of
Indian Logic, p. 299, fn. 2.
73 . The doctrine of Hew caka is also mentioned i n Nta Pavda.
74. In the Ntya Vtrtika of Uddyotakara in the context of examination of the doctrine
of trairea there is a statement of another doctrine of seven possible types of relations
hew may have with pala, sapala, and asapala. The seven possibilities are:
1 . present in pala, absent in sapala, and present in asapala.
2. present i n sapala, absent i n pala, and present i n asapaka.
3 . absent in asapala, absent in pala, and absent in sapala.
4. present i n pala, present i n sapala, but not absent in asapala.
5 . present i n pala, absent in asapala, but not present i n sapaka.
6. present i n sapala, absent i n asapala, but not present i n paka.
7. present i n pala, present in sapala, and absent in asapala.
Of these the first three fail to satisf two requirements, the next three fail to satisf
one requirement, but te last one satisfes all three requirements. Thus, the frst six are
rej ected, and the seventh alone is regarded as valid. The hetll in the seventh placement is
traireipa hew. This septenary is diferent fom te Hew caka insofar as it takes into
consideration the relation of hetll with pala apart from its relation with sapala and
asapala. Here the possibility of asiddha hetu is also taken into consideration. But te
doctrine hew cakr presumes that hetu must necessarily be present in pala. With this
presumption it considers nine possible ways of te relation of the hew with sapala and
asapaka.
75 . Borrowed fom Chi , Buddhist Formal Logic.
76. Vacaspati, Nta Vtrtik Tttpara TiM, p. 266. Avaavatvenaikavakatt darsitt,
sa ca padtndm parsparapelita sambandha yogyarthaprtaena bhavanti.
77. Vatsyayana, Nta Bhdra 1. 1 . 32.
78. Ibi d. , I . 1 . 1 .
A
gamal pratiia hetllranumanam udthar/am pratakam
lIpamtna upanayal. SCvetmektrtha samavtye stmarthya pradarsanam
nigamanamiti. So 'yam parm raya iti.
79. Ibid. , 1. 1 . 32.
80. Partasarati Misra, Ntya Rtna Mala on
S
lok Vtrtik, anlllnana khala,
p. 54.
1 1 6 Notes
81 . Vacaspati, Niya Virtika Titpara Jki, p. 267.
82. Viptj viii!a pa!adharmatijfinam partmariam.
83. In te Buddhist literature prior to Diimaga there is an acceptace of eight avavas,
which have only been stated and not analyzed (cf. Tucci, Buddhist Logic before Dilntga,
pp. 461-65) . In the Naa Pavea, however, only three avaavas, namely, prtijit, hetu,
and dr!tnta, are put forth (pp. 1 -2) .
8 4.
S
ataralita, Tata SaJlgra, verse 1435.
85. According to them, in partrthtnumtna we have five avayavas, but i n
svtrthtnumtna three are enough.
86. Quoted in
S
antarak$ita, Tatta Samgrha, pp. 514f.
87. Ibid.
88. Ibid . , p. 516.
89. Satkari Mookerjee, Buddhist Philosoph of Universal Fll, p. 364.
90. Diiaga, Ntya Prveia, Tr. by Dhruva.
91 . Kamalasila, TatNa Samgrha Pafjikt, p. 51 6.
92. Dhannaklrti , Prmtla Vtrtik; Ch. 4, verses 1 6, 1 7, 20, and 26.
93. Dharmaklrti , Pramtla Virtika, Ch. 3 , verse 27 . Tadbhtvahetubhtva hi
dr!tnte tadvedinal.
94. DharmakIrti, Naa Bindu, Ch. 2, p. 11 7.
APPENDIX 2
1 . According to Buddhist tradition, the purpose of acquiring knowledge is either for
one' s own awareness or for communicating it to others. Buddha acquired knowledge for
his own enlightenment and also communicated knowledge for the enlightenment of others.
So, Buddhist epistemological thinkers draw a distinction between knowledge for one' s
own sake (ttmasamvit) and knowledge for the sake of others (parsal1ivit) . Perception
(prta!a) and inference (anwni na) are the two means to acquire knowledge for one' s
own sake. Anlllnina in this context is svtrthtnlllntna only. For communicating knowledge
to others there is only one mode, which is sidhana, also known as partrthinumtna. In
this text first there is exposition of sidhana, and then prta!a and anumtna are discussed
with their respective fallacies. So, in total, there are eight topics discussed in this text.
They are as follows: ( 1 ) Sidhana, (2) Stdhantbhtsa, (3) Duala (4) Daaltbhtsa, (5)
Patya!a (6) Pata!tbMsa (7) Anwnina (8) AnultntbMsa.
In a proper epistemological order first there should have been di scussion on
perception and inference and then on stana and dial}a. Bu the order i s reversed here.
The tradition explains it by saying that this book is intended for the sake of others
(parsmivit) and therefore ttmasmvit is made subservient to parsmilvit. Buddha also
acquired enlightenment not for his own sake primarily but for sufering beings (Bahujana
hitaa bahujana sukhaa) .
The word "stdhana" is used in two di fferent senses . One is the mode of
demonstration or giving arguments in support of one' s own position, and the other is for
one of the constituents of the process of inference, that is, hetu (reason) . Here, it is used in
the former sense. The entire process of argumentation consisting of different elements and
stages is collectively known as stdhana. The elements are thesis (a!a) , reason (hew),
homologue (sapa!a), and heterologue (vipa!a) . The stages are thesis, reason, ad example
(drttnta) . Pointing out fllacies in stdhana is known as daala. Daala means pointing
a fallacious siana, that is stdhantbMsa. It means pointing out such defects in stdhana
that are existent here. Daala is directed against stdhantblisa and can never be directed
Notes 1 1 7
against stdhana. For a genuine stdhana there cannot be any daaJa. Only when the
stdhana is fallacious is there a possibility of diaJa. In fact, stdhana and daala are the
two positions adopted by the two parties in a debate. The proponent puts forth one's
positio
n
through sidhana, and the opponent tries to refute it through daaJa. A fallacious
stdlana is known as
s
tdhantbhtsa. A fallac may be because of defect in thesis or
reason or example. To catch the fallacious stdhana, that is, to point out stdhantbhtsa, is
known as daaJl a.
.
Daalabhtsa means pointing out defects in a sihana that are nonexistent. It is
allegating defects in a stana that are not there. Stdhana and daaJa, along with their
respective fallacies, are meant for communicating knowledge to others. All these four are
involved in a debate where one party tries to refte the other. These are, therefore, diferent
aspects of vida-vidli (ar of debate) .
Perception and inference are the two modes of acquiring knowledge. Buddhists
belonging to the Di imaga-Dharmakiri tradition accept only two prmilas, namely,
perception and inference. A knowledge is acquired by a knower for one's ow
n
sake.
According to Buddhists, perceptual knowledge concers te objective real kown as svalal
aJa. A svalalaJla is devoid of all determinations and characteriztions; therefore, a
perceptual knowledge is aso indeterminate. In this respect Buddhists fndamentally difer
fom Naiyayikas, according to whom a perceptual knowledge can be determinate as well .
Buddhists accept five types of determinations (kalpant) , namely, in terms of substance,
qualities, action, class-character, and language. Buddhists insist that perception is devoid
of all determinations, and, therefore, they defne perception as kalpantpotha. Knowledge
of an object by means of a necessary mark is known as anUntna. A distinction is drawn
between svtrthtnltl1ltna and parttltnllllltna. Svtrtlnlllntna is just cognitive
(ftnttmak) , whereas partrthtnumtna is expressed in verbal propositions (iabdttmaka) .
Parithtnumtna is covered up under stdhana, and under anumtna as a mode of knowing
svtrthtnlllntna is accepted. There is a possibilit of defects in both these modes of knowing.
The defective perception is known as prtaltbltsa, and defective anumtna is known as
anllmtntbhtsa. Perceptual knowledge i s necessarily indeterminate. But if i t i s
misunderstood as determinate knowledge, then i t is a fallacious perception. Perceptual
knowledge concers objective reality (svalalaJa) . A svalakaJa can only be grasped
(gra) and not conceptalized (adlravasita) . But if in a perceptual kowledge te objective
reality is conceptualized, then it becomes prtaktbhtsa.
Anllmtntbhtsa is fallacious inference on account of fallacious reason (hetvtbhtsa) .
Since anrlllltna is taken here as svtrthtnllmtna, only Ietvtbltsa is mentioned, not prtak
tbhtsa and dr!tnttbhtsa, which come under stdhantbhtsa.
This is the synoptic account of the vtda-vidhi (theory of debate) and prmtJa-vidhi
(theory of knowledge) , which are the two aspects of nia concering knowledge. There
are works dealing exclusively either with prmaJla or with vida, but here both are discussed
together. The verse tat occurs at the beginning of the text sums up the entire discussion
on vtda and prmtJla presented in this text, and therefore it prefaces the text.
2. Stdhana consists of three elements. They are pala, hetlt, and dr!tnt. A stdhana
is always employed to establish something that is not known to the other party. Therefore,
it is necessarily communicative (vacana) . In fact, it is employment of parttlztnlllntna in
debate. When svarthanllmana is verbalized, it becomes pardthanllmana, and when
parithanlllntna is used in debate, it is known as sadlzana.
3 . Paka is that which is brought to consciousness or manifested (acate iti pala() .
Pala is always a subject taken with its predicate. It has two predicates. One is known or
1 1 8 Notes
wellknown, whereas the other is yet to be known. The latter is to be known on the basis of
the former. According to Buddhists, te pala is the object of inferential inquiry and
subj ect of proof. The defini ti on of paka in the Buddhist tradi ti on as svayam
stdatvenepsitaJ means that it is someting that one desires to prove; tat is, it is something
that is upheld by the proponent (vtdi) and directed to the opponent (prtivtdi) for proving.
Reason is the means through which the hitherto not proved is proved .
Etmologically, hetu means hi. hinoti, to know, which means that by which something is
proved. The pala and te hetu should be accepted by both the paries; only then can they
make the hitherto unproved, proved.
4. These three features of reason constitute the foral conditions of the validity of
anumtna and therefore of stdhana. Violation of any of these three results in a fallacy.
Since there are three features of reason, terefore, there are tree types of fallacy peraining
to reason.
5 . Homologue and heterologue are the supporting factors for proving a thesis.
Homologue is affirmative, and heterologue is negative in natre. Both are grounded in
experience, and their veracity is wellestablished. Homologues are known substrata of the
presence of probandum, and heterologues are the known substrata of the absence of
probandum. The homologue provides positive support, and te heterologue provides
negative support to reason in proving the probandum. Since the relation of probandum
with the substrata involved in homologues and heterologues is that of logical necessity,
both of them provide the logical ground for the relation of invariable concomitance beteen
reason and probandum. The homologue points out that there is at least one such case that
is similar to subject and that can serve as a corroborative instance for the existence of
probandum in the particular subject where reason is also present. The heterologue also
provides a negative support insofar as it points out that there cannot be any presence of
reason in the absence of the presence of probandum, and, therefore, the presence of
reason in the particular subject provides the necessary ground for the presence of probandum
in that particular subject.
6. Enumeration of exmple provides frther support to the reason in proving its
probandum. Since te relation between te reason and the probandum can be either positive
or negative, accordingly, there can be to types of example. The positive example instantiates
the homologue, and the negative example instantiates the heterologue. The diference
beteen heterologue and homologue and their corresponding examples can be understood
in terms of a formal statement of a relation and its concrete instantiation. Historically,
before the idea of generalization or universalization came to be formulated, examples
alone served as the basis for proof. But when the need and necessity of generalization
were met in terms of vapti or the "relation of invariable concomitance, " the role of
example became subservient to that of vtpti . Nevertheless, te probative role of the later
was not discarded or belittled. The sadhara dr!anta is of the same natre as that of the
sapala. Likewise, vaidhanra dr!anta is of the same nature as that of the vipala.
In the Buddhist tradition sadhana. which stands for a means to communicate
knowledge to others, aways consists of three elements, the thesis, the reason, and the
example. All three have to play their respective role in the stdhana
.
In other words, the
thesis must be wellformulated, the probative reason must be welladduced, and the
corroborative examples must be suitably enumerated.
7. In the formulation of a stdhana if any of the three constituents are vitiated by some
defects, then it is known as stdhantblsa. The stanablsa can be of three types: pak
tblsa, hettbhtsa. and dr!tntablsa. In the Nyaya system the roles of pala and d
rH
Notes 1 1 9
lnta are made subservient to the role of hem, and therefore it concentrates on hetvlbhlsas
alone, but Buddhist logicians take into account te other two types of fallacy also. Let us
clarify some points regarding these nine pallbhdsas one by one.
(i) Since we know by perception that words are audible (or there is audibility in words) , the
statement "Words are inaudible" is contradicted by perception,
(ii) It is contradicted by the inference " A pot is noneternal because it is a product" . .
(iii) In Vaiseika metaphysics sound is regarded to be produced, and, therefore, it is noneternal.
(iv) Acording to popular belief, the former is regarded as untouchable, though not the latter.
(v) It is not clear here whether a statement that contradicts one's former statement will also
come under this head.
(vi) Acording to Buddhists, everything is momentary and therefore noneternal; but according to
satkdravdda of Sirkha, all entities preexist in their cause, and, therefore, they cannot
agree to sabda being perishable.
(vii) Here, the existence of self is denied by Buddhists. They deny both the subject and the
predicate.
(viii) They don't believe in the existence of soul or hold that pleasure and so on are qualities
inlering in soul.
(ix) ' Sound i s audible" i s a fallacious thesis because something is desired to be proved
(stdhyatvellepsita/l) only when its truth is at issue between the two parties, but this is not the
case with "Sound is audible" , and so it is a fallacious thesis. This is a fallac of triviality
because it is absurd and fallacious to offer as a thesis a statement that everyone would accept
as a plain statement of fact .
8. According to the doctrine of trirapa there are three conditions which regulate
the validity of a reason (hew) . Violation of any of the three conditions results in making
te reason invalid. Thus, tere are three types of fallacies of reason.
9. Sapalaikefavrtil vipalavapi : Inconclusive (anekintika) is sixfold. The first
two, being easy to understand, do not require explaination. Here we are giving te detailed
explanation for the remaining four.
Thesi s: Sound is naturally existing (non efort-made)
Reason: Noneternalit
Homologoue: Lighting, space, and so on
In one part of homologue, that is, lightning, reason (noneternality) resides but not in
te other part, that is space.
Thesis: Sound is natrally existing (non effort-made)
Reason: Noneterality
Heterologue: Pot
Reason (noneterality) resides in the entire class ofheterologues; therefore, the reason
is inconclusive here.
10. Vipalaikdefav.rtti sapakavapi:
Thesis : Sound is efort-made
Reason: Noneterality
Homologue: Pot, and so on
Here, reason resides in the entire class of homologues.
Thesis: Sound is effort-made
Reason: N oneterality
Heterologue: Lightening, space, and so on
Here, reason resides in one part of the class of heterologues, that is, lightning and so
on, and doesn' t reside in another part, that is space and so on. Therefore, the reason is
120
inconclusive here.
11 . Ubhayapaaikadeia v
.
rfi:
Thesis : Sound is eteral
Reason: Incorporeality
Homologue: Space, infnitesimal particles, and so on
Notes
Here, reason resides in one part of the homologue, that is space and so on and does
not reside in another part , that is infinitesimal particles and so on.
Thesis: Sound is eternal
Reason: Incorporeality
Heterologue: Pot, pleasure and so on
Here, reason (incororeality) resides in one part of te heterologue, tat is, pleasure
and so on, and does not reside in another part, that is, pot and so on. Therefore the reason
is inconclusive here.
12. Viruddhavabhicarf:
Here, two diferent theses are given, one put fort by Vaiseikas and anoter by
Mlmamsakas. According to Vaiseikas "Sound is noneternal because it is produced, like a
pot and so on" , and, on te contrary, according to Mlmasakas, "Sound is eteral because
of its audibility, like " soundness. "
(a) Thesis: Sound is noneternal
Reason: Produced
(b) Thesis: Sound is eternal
Reason: Audibility
Here, we have two contradictory theses supported by two different valid reasons
about the same subject, sound. A Vidyabhushana puts it, "This fallacy arises when a
tesis and its contradictory are both supported by what appear to be valid reasons. " Tis
fallacy is known as viruddMvabhicarf because the theses are mutually contradictory, but
they are validly established by their respective reasons. If taken separately, each appears to
be valid, but doubt arises only when they are taken together. Here it can be asked how this
fallacy is not diferent fom te fallacy of "too narrow" discussed earlier. The reply is tat
it arises only when. one argument alone is taken independently of the other. But when both
are taken togeter, it would be the fallacy of "viruddMvabhicarl" .
In the Naya Bindu Tka Dharmottara gives two possible meanings of the term
"viruddMvabhicari ":
1. avabhicari ofviruddha, that is, that which invariably proves the opposite of what
is proved by another hew.
2. viruddha, which is at the same time avabhicari, that is, that which is opposed to
anoter hetl inasmuch as it proves its contradictory and is at te same time invariably
accompanied by its own sadlra, that is, which is good in itself, so far as its own vapti is
concered.
In the example " Sound is noneternal on account of being produced like a pot , "
"being produced" is opposed to, or contradicted by, "Sound is eteral on account of
audibility like soundness. " The two reasons taken together cause doubt and are inconclusive.
The fnction of a reason is to prove the existence of te probandum in a particular
locus or paa. A reason can perform this fnction if and only if it stands in the relation
of necessary concomitance with its probandum. This implies that the reason should not be
such as to prove the opposite of either the probandum or the thesis concerning the
Notes 1 21
probandum. Likewise, i t should not disprove the very existence of the probandum or the
thesis, nor should it disprove any essential property of the probandum or the thesis.
Accordingly, four varieties can be conceived of the con
t
radictory reason, namely, pertaining
to reason or to a particular reason and pertaining to a thesis or to a particular thesis. One
type of fallacy is generi, and the other one is specifc or particular; one covers up the
entire existence, whereas the other covers up only a particular property.
1 3 . Dharmasvarupaviparita sadhana:
Thesis: Sound is eteral
Probandum: Eternal
Reason: Produced / effort made
If anything is produced, it can' t be eteral . Likewise, if anything is efort made, it
must be naturally existing, and, therefore, must have a beginning. These two reasons
don' t reside in eteral entities (homologues) like soud; rather, they reside only in noneteral
entities (heterologues) . Thus, instead of proving eteralit of sound, they would prove the
opposite of eterality.
14. Dhar/laviea viparftasddhana:
In the argument "Eye and so on exist for the sake of others (i . e. , self on account of
their being of the nature of aggregates like bed, seat, and so on, which have parts" , here,
we have,
Thesis:
Reason:
Probandum:
Eyes and so on exist for the sake of others
Aggregateness
To exist for the sake of others
This argument not only proves te existence of eyes and so on for the sake of others
on the ground of being aggregates, but also proves the aggregateness of the "self" . This is
because if eyes and so on are aggregate, then one whose purpose they serve should also be
aggregate. This is because of transference of property. However, this is not acceptable to
the Samkhya school, which is the opponent here because, according to the Sarya school ,
self is not an aggregate. In this argument the reason is fallacious because it proves the
opposite of the particular dhamza or property that it is intended to prove. The reason, that
is, aggregateness, while proving that there is something beyond eyes, proves also that
something, namely, tan, which Sarkya desires to prove, for the same reason is proved
to be having aggregation, a position that Sarkhya is not prepared to hold. Sarkya
thinkers on the basis ofthe previously stated reason want to prove not only the existence of
the individual self but also its being noncomposite. But, this argument not only proves the
existence of self but also proves its composite character.
1 5. Dharmisvaripaviparftasddhana:
The example given here is "Existence (bhdva) is not a substance or action or quality,
because of its possessing one substance and because of its presence in qualities and actions
like the universal and particular."
Thesis
Probandum
Existence (mahasdmdnya) is slllnmllln-genlls
Existence is diferent from substance or
action or quality
But this argument is vitiated by a falacy of hem contradicting a tesis, say Buddhists.
Because the same bipartite hetu, namely,
Reason: (a) Because it possesses a single substance
(b) Because of its presence in qualities or actions,
would prove the absence of the thesis, that is, bhava as well as absence, that is, abhava.
Thus, this bipartite hetu may prove that bhdva is abhdva, which is the very opposite of the
122 Notes
thesis. Buddhists here are refting the Vaisesika concept of blva, which is the summum
genlls, on the ground that the reason put forh to establish the existence of bhtva, in fact,
disproves its very existence. It no doubt proves that existence is different from substance,
qualities, and actions, but it also proves the absence of existence.
1 6. Dhannivisea viparUastdhana:
Here, the example given is the same as before with the addition that the thesis is taken
along with its particular property, namely, the causal capability of giving rise to the notion
"It is existent" . In this argument both the reasons not only prove the property of causing
the notion "i t is existent" but also prove the absence of this property. The VaiSesikas are
realists in their epistemology, and therefore they argue that existence aone is the cause of
the notion that " it is existent . "
1 7 . In the theory of inference, example plays a signifcant probative role. I t acts as a
support to the reason in terms of a concrete exemplifcation. It enumerates a case similar
to the vtpt situation, that is, a locus in which reason necessarily coexists with the
probandum. Such cases are known as sapala or homologue. A sapala is always similar
to pala insofar as it possesses the coexistence of reason and probandum. But in terms of
assuredness of the coexistence of reason and probandum, it is on a strong footing compared
to that of pala. An example based on similarity reinforces the positive concomitance, and
an example based on dissimilarity reinforces negative concomitance. So, the former consists
of enumeration of homologue, and the latter consists of enumeration of heterologue. If the
example is not in conformity with homologue or heterologue, it is fallacious. Thus, there
are two types of fallacy of example: one in which sapala is not valid, and the other in
which vipaka is not valid.
1 8 . Stdhana dhanntsiddha:
The example given is "Sound is eternal on account of its being incorporeal like
infinitesimal particle. " Here,
Thesis: Sound
Reason: Incorporeality
Probandum: Eterality
Homologue: Infinitesimal panicle
Here the homologue (i . e. , infinitesimal particle) no doubt possesses the probandum
(i . e. , eternality) but doesn' t possess the reason (i . e. , incorporeality) . Therefore, the
homologue is not similar to the thesis. The reason does not exist in the homologue, and
therefore the homologue does not have any probative value. Only that homologue has
probative value that has the presence of both the reason and the probandum.
1 9. Stdlradhartsiddha:
In the exmple "Sound is eteral on account of its being incororeal like intelligence" ,
Thesis : Sound
Probandum: Eternalit
Reason: Incorporeality
Homologue: Intellect
Here the homologue (i . e. , intellect) no doubt possesses the reason (i . e. , incorporeality)
but does not possess the probandum (i . e. , eterality) . Therefore, the homologue is not
similar to the thesi s. As before, the homologue does not have the probative value because
it lacks the probandum.
20. Uhaadhanntsiddha:
This is twofold: afrmative and negative. An exmple of the fonner is "Sound is
eternal on account of being incorporeal like a pot" . Here,
Notes
Thesis : Sound
Reason: Incorporeality
Probadum: Eterality
Hqmologue: Pot
123
The homologue (i . e. , pot) has absence of both reason and probandum; therefore, the
homologue is not similar to the thesis and therefore has no probative worth.
A exaple of the latter is "Sound is eteral on account of being incororeal like
space" . (This argument is valid only for those who believe in the reality of space, but it
will not be valid for Buddhists, who do not believe in the reality of space. ) Here,
Thesis : Sound
Reason:
Probandum:
Homologue:
The existence of homologue (i . e. ,
Buddhists) .
21 . Ananvaya:
Incororeality
Eternality
Space
space) is not acceptable to one party ( i . e. ,
The statement of invaiable concomitace is te very backbone and logical foundation
of inference. No inference can be valid in the absence of such a statement. An example is
attached to such a statement. If an example is given without a statement of invariable
concomitance, then it cannot have frm probative worth. Such an example will be fallacious.
In such a situation the mere coexistence of probandum and reason is stated without pointing
out their relationship of invariable concomitance.
22. Viparitanvaya:
In the argument in the pot there are the properties of "producedness " and
" non eterality" , the example lacks positive invariable concomitance. It would be a valid
example if it is restated as "Whatever is produced is noneternal like a pot. " This fallacy
occurs when a positive invariable concomitance is stated inversely and therefore wrongly.
Thus, instead of saying, "Whatever is produced is noneternal , " if it is said that "Whatever
is noneteral is produced, " then the order of reason and the probandum are reversed, but
thereby the pervader becomes the pervaded, and the pervaded becomes the pervader.
23 . In dissimilarity we have a statement of heterologue that should have the absence
of both reason and probandum. Like the fallacies based on similarity these are also of five
types. I the classification of the fallacies based on similarity, the starting point is reason
because the presence of reason implies the presence of probandum, whereas in the
classification based on dissimilarity, the absence of probandum implies te absence of
reason.
24. Savavrta:
The example " Sound is eternal on account of incorporeality l ike infinitesimal
particle" , if converted to a statement of negative invariable concomitance containing
dissimilar example of heterologue, would then rn as: "Whatever is noneternal is corporeal
like infnitesimal particle. " As per the rule, the same example canot be adduced bot as
heterologue and homologue. But in this case, the same example, that is, infinitesimal
particle, is put forth in both situations. Just as in the invariable concomitance between
smoke and fre, where kitchen can serve as a homologue, but it cannot serve a a heterologue,
whereas lake can serve as a heterologue, likewise, in this case also infinitesimal particle
can serve only as a homologue and cannot serve as a heterologue. In infnitesimal particle
incororeality, which is reason, is absent, but noneterality, which is probandum, is not
absent, whereas as per the rule both should be absent. Thus, it contains the fallacy of
124
nonabsent probandum.
25 . Sadhanavavrtta:
Notes
Here, the thesis is same as before, but the example i s different. Instead of
infinitesimal particle here we have action. In this argument, the statement of positive
invariable concomitance, "Whatever is incorporeal is eteral" , if converted to a statement
of negative invariable concomitance containing dissimilar example of heterologue, would
then run as, "Whatever is noneteral is corporeal like action" . Al explained before, the
same example cannot serve as both homologue and heterologue. In "action" the probandum,
that is, eterality, is excluded, but the reason, that is, incorporeality, is not excluded.
26. Ublaavavrtta:
Again, the thesis is the same as before but the example is different. Instead of an
infinitesimal particle or action, we have space here. In this argument the statement of
positive invariable concomitance, "Whatever is corporeal is eteral , " if converted to a
statement of negative invariable concomitance containing dissimilar example or heterologue,
would then run as, "Whatever is noneterna is corporeal like space" . Here, both reason
and probandum are not absent because space is both eteral and incorporeal .
27. Avatirek:
Only the absence of probandum and reason is said to belong to the exmple without
stating their negative invariable concomitance. Thus, instead of saying, "Whatever is
noneteral is not incororeal like a pot, " if we simply say, "In pot reside corporeality and
noneterality" , then it is a case of nonstatement of negative invariable concomitance.
28 . Viparita vatireka:
While stating negative invariable concomitan
c
e, first the absence of probandum is
stated, and thereafter the absence of reason is stated. But here, first the absence of reason
is stated, and thereafer absence of probandum is stated. This is the fallacy of reversed
absences because here the logical order of absences is not followed. Only absence of
probandum implies absence of reason, not vice versa (sadabhtve sadhanabhtva/) .
The treatment of the fallacious means of proof comes to an end with the exposition of
its tree varieties. The autor discusses te means of knowledge tat are helpfl in acquiring
knowledge for oneself. Thus, svasmnvit and atmaprtayana are synonyms, as parsmnvit
and parprtaana are synonyms.
According to Buddhists, the theory of knowledge is dependent on the theory of realit.
Since there are two and only two types of objects, there are two and only two modes of
knowing. One and the same mode of knowing cannot cognize both types of objects. Thus,
for Buddhists perception and inference are the only two ways of knowing. It is not that
Buddhists deny other ways of knowing accepted in other schools. They only deny their
independent status and reduce them to inference.
Glossa
1. Abhtsa: Fallac
2. Abhata stdhanadoa: Nonexistent defect in means of proof
Contrary to one's own scripture
Absence of statement of invariable
concomitance
3 . Agamavirddha:
4. Ananvaa:
5.
6.
7.
8.
9.
10.
1.
12.
1 3.
1 4.
1 5.
1 6.
1 7.
1 8.
1 9.
20.
21 .
22.
23 .
24.
Anektntika: Inconclusive
Anumtna: Inference
Anumtntbhtsa: Fallacious inference
Anumtnavirddha: Contrary to inference
Anatartsiddha!: Unacceptable to either
AprsiddhaviseaJa: A tesi whose subjet is uaccetble
Aprsiddhavisea: A tesis whose prcate is uc tble
Apraiddhobhaa: Unacceptable to bot parties
Apryattnantarlyaka: Natral, noneffort-made
Astdhtrala: Too narrow
Asiddha: Unproved
Avidyamtna: Absent
Avatireka: Absence of negative concomitance
Asratsiddha: Fallac of unproved locus
Dharr1 pavpartMa: Rason contrary to probandum
Dharaviseaviparlastdhana: Reason contrary to a particular probandum
Dharisvar1 pavipartMa: Rason contrary to thesis
Dhariviseaviparlastdhana: Reason contrary to a particular tesis
Doa: Defect
Dr!tnta: Example
126
25. Dr$!tnta-do$a:
26. DU$aa:
27. DU$a1bhtsa:
28. Hetu:
29. Hetvtbhta:
30. Kalant:
31 . Linga:
32. Lkavirdha:
33. Paka:
34. Paktbhta pratiftdo$a:
35. Pakadharatt:
36. Pakadharatva:
37. Parmilu:
38. Psiddha Sambandha:
39. Pika:
40. Prtak$a:
41 . Pratak$tbhtsa:
42. Patakavirddha:
43. Praattnantarfyaka:
44. Stdhana:
45. Stdhtntbhta:
46. Stdhanahartsiddha:
47. Stdhanado$a:
48. Stdhantvtvrt i :
49. Stdhtrala:
50. Stdhana:
51 . Stdhana dr$!tnttbhtsa:
52. Stdha:
53 . Stdhadhartsiddha:
54. Stdhtvtvrti :
55. Stmtnalak$a1a:
56. Sapaka:
57. Sapak$aikadeiavrtti :
58. Sapak$avtpi:
59. Svalak$a1a:
60. Svavacanavirddha:
61 . Uhaadhartsiddha:
Gl os s ary
Fallac in example, fllacious example
Fla i meas of pror rftton
Fallacious reftation
Reason
Fallac in reason, fllacious reason
Conceptualization
ecessary mark
Contrary to public opinion
Subject, tesis, locus
Fallacious tesis
ecessary presence of reason in the
thesis
Thesis h rn a it prrt
Infnitesimal particle
A thesis too well known or accepted
universally
Inquirer
Perception
Fallacious perception
Contrary to perception
Efort made
Means of proof
Fallac in te means of proof
Incompatibility of the exampl e with
the reason
Defectiye means of proof
onabsent reason
Too wide
Similarity
Fal lacy of exampl e on the bas i s of
similarity
Probandum
Incompatibil ity of the exampl e with
te probandum
onabsent probandum
Generalised image
Homologue
Pervading one part of homologue
Pervading entire homologue
Unique particular
Contrary to one's own statement
Incompatibility of the exampl e with
bot (reason and probandum)
Gl os s ary
62. Uhaapaaikadesavrt i :
63. Uhaisiddha:
64. Uhayavtvrta:
65. Uhaytvrta:
66. Vaidhara (vatireka) :
67. Vaidhara dTtnttbhtsa:
68. Vidyamtna:
69. Vikalpa:
70. Vpaa (Asapaa) :
71 . VipaaikadeSav.rtt:
72. Vipak$avtpi:
73. Viparitnvaya:
74. Vipariavatireka:
75. Virddha:
76. Virddhavabhictri:
77. Vi$atktra (arhtktr) :
Pervading one part of bot
Unacceptable to bot
Bot are nonabsent
Bot are absent
Dissimilarit
127
Fallac of exmple based on dissimilarity
Present
Concept
Heterologue
Pervading one part of te heterologue
Pervading entire heterologue
Contrary to the statement of positive
concomitance
Contrary to te statement of negative
concomitance
Contradictory
Contradictory and divergent
Form of the object
Bibliogaphy
Aniruddha, AbhidhammatasGlngaho . Traslated by Shwe Zan Aung and C. A. F Rys
Davids, London, Pali Text Soceity, 1910.
Aalga, MahtyanasrtralGkar. Edited by Vaidya P. L. , Darbhanga, Mithila Institute,
1 960-64.
Akalaika, Laghfyastrya. Edited by Mahendra Kumar, Ahmedabad, Sindhi Jain granth
Mala, 1 939.
Arcata, Hew Bindu T. Baroda, Gaekwad Orienta Institute, 1 949.
Bhartrhari, Vakapadfya. Edited by Pillai K. R. , Delhi , Motial Baarsidass, 1 971 .
Bhatt, G. P. , Basic Was of Knowing, Delhi , MotHal Baarsidass, 1 989.
Bhatta, Jayanta, Nta Mafjari. Edited by Surya Narayana Shukla, Varanasi , 1 971 .
Bhatta, Kumarila,
S
loka Vartika. Darbhanga, K. S. Darbhanga Sanskrit Universit, 1 979.
Chatterj i , Durgacharana, "HetucakrniTaya, " Indian Historical Quarterly 9, 1933, pp.
266-272,
~
. "Sources of Buddhist Logic" (From the Traditional Point of View) . Indian Historical
Quarterly 9, 1 933 , pp. 499-02.
-. "The Doctrine of Trirupa Hetu. " Indian Historical Quarterly 9 , 1 933,
pp. 503-10.
~
. "Sources of Kowledge i n Buddhist Logic. " Indian Culture, Calcutta, no. 2- Oct.
1934 pp. 263-74.
. "Hetucakrhamaru" (also known as Hetucakrnirfaya or Hetucakr{amaru)
(Translation into English) , Indian Historical Quarterly, 9, 1 93
'
3 , pp. 511 - 1 4.
Chatterj i , S. c. , Nta Theor ofKnowledge. Calcutta, Calcutta University, 1939.
Chi , Richard S. Yo , Buddhist Formal Logic (A Study of Diimaga' s Hetllcaka and K' Uei
Chi 's Great Commentary on te Nta Pavda) . Delhi , MotHal Banarsidass, 1984.
Chinchore, Mangala R, Vadaraya; A Glimpse ofNta-Buddhist Controvers, Delhi ,
Sri Sadguru Publications, 1 988.
-. "Some Thoughts on Significant Contribution of Buddhist Logicians. " Joural of
Indian Philosophy, 1 5, no. 2 June 1987, pp. 155-71 .
Croce, Bendetto, "On Indian Logic. " East and West (Roma) , 4, 1 953, pp. I-30.
Das, Rsvihary, "Relations in Modern Indian Logic. " The Calcutta Revie, 45, no. 2,
1932, pp. 143-60.
Dasgupta, S. N. , A Histor ofIndian Philosoph, 5 vols. Cambridge, Cabridge University
Press, 1 951 .
.
Dharati . Hew Bindu (wit Tk by Arcata and
A
loka by Durveka Misra) . Edited by
1 30 Bi bl i ography
Saghavi , Sukhalalj i , ad Muni Jinavinayaj i , Baroda, Baroda Oriental Institute,
1 949.
. Naa Bindel. Edited by Shastri, Chandrasekara, Banaras, 1 954.
. Prmtfa Vtrtika. Edited by Pandeya R. C. , Delhi , Motilal Banarsidass, 1 989.
. Santtntntarsiddhi . Edited by Kitagawa and Hidenoir. Joural of Greater India
Societ, 1 4, no. 1 , 2, 1 955, pp. 407-29.
. Svtrthtnlllntna-pariccheda. Edited by Dalasukha Bhai Malvaiya, Varanasi , 1 959.
. Vtda Naa (with Viparicittrtha by
S
antarakita) . Edited by Sankrityayana, Rahula.
Appendix to Joural ofBihar and Orissa Research Societ, 1 949, pp. 21 -22.
. Vtda Naa (with by Vipafcittrtha and Smibandhaparf!t) . Edited by Swami
Dwarikadas Shastri , Varanasi , 1 972.
. PrllltJa Viniscaa. Edited by Steinkellner, Wien, Verlag Der Ostereichischen
Aademe Der Vissenschafen, 1973 .
Dharmottara, Naya Bindll TM. Edited by Shastri Chandrasekhar, Banaras, Chaukhamba,
1 954.
Dharmaraja Adhvarlndra, Vedtnta Paribhtt. Translated by Swami Madhvananda, Belur
Math, Rama Krishna Mission, 4t edition, 1 972.
Diimaga.
A
lambana Parf!t Vnti. Edited by N. Shastri Aiyaswami, Madras, Ayar Library,
1 942.
. Naa MllkJa, Edited by Tucci Guiseppe, vol . 1 . Heidelberg, Jabllch des Institutes
fur BuddhisllllS Kunde, 1930.
Ntya Prvda. Edited by B. J. Sandesara, Baroda, Baroda Oriental Institute,
1968.
---. Pramtfa Samuccaya (Pratyaka Khafra) ( Svavrttisahi ta) . Edi t ed by
H. R. Iyengar Rangaswamy, Mysore, Mysore Press, 1930.
Durveka, Misra, Dharmottar Padfpa. Edited by Dalsukhabhai Malvania, Patna, Kashi
Prasad Jayaswal Research Institute, 1 971 .
Gokhale, P. P. , Inference and Fallacies Discussed in Ancient Indian Logic. Delhi , Sri
Satguru Publications, 1 973 .
Gupt a, Praj iakar a, Pramtfavtrtiktlmiktra (al so known as Pramtfavtrtika
Bhta) . Edited by Sankrityayaa, Rahul, Patna, Kashi Prasad Jayaswal Research
Institute, 1953.
Gupta, Rita, "The Buddhist Doctrine of Momentariness and Its Presuppositions. " Joural
of Indian Philosoph 8, no. 1 , March 1980.
Hastings, James, Encclopaedia ofReligion and Ethics. Edinburah: T. & T Clark, New
York.
Hattori , Masaki . Difmaga on Perception (Patya!a Pariccheda of Diiaga' s Pamafa
Sailluccaya) . Cambridge, 1968.
. "Fragments from Prmafa Samllccaa. " Joural ofIndian Buddhist Studies, no.
1 , 1958, pp. 330-35 .
. "Pata!abhdsa" (Diaga's View ad Dhaakri ' s Iterpretation), Miscellanea
Indologica Kiotiensa, 6-7, March 1 965, pp. 122-28.
Hayes, Richard P. , "Diiaga' s Views on Reasoning" (First half of Svarthtnltmana
Pariccheda ofPrmtfa Samuccaa) , translated into English in Journal ofIndian
Philosoph, 3 , 1980, pp. 21 9-77.
Hemchandra, Pramtfa Mfmtmst. Varanasi , Tara Publications, 1 970.
Herzberger, H. G. , "Double Negation in Buddhist Logic. " Joural ofIndian Philosoph,
3 , March/April 1 975 , pp. 3- 1 6.
Bibliography 1 31
Isvara KrJa, Samka Karik. Edited by Chaturvedi Brij Mohan, Delhi, National
Publishing House, 1970.
Iyengar, Rangaswami
. "Some Teories of Buddhist Logic in the KavyalaIkara of Bhamaha. " Proceedings
of All India Oriental Conference, No. 8, 1935, Mysore, pp. 419-25.
. "The Voa Vidhi and the Voa Vdhina of Vasubandhu. " Brhmavidya. Te
Adyar Librr Bulletin 17, part I, 1953, pp. 9- 19.
J ayatilae, K. N. , Earl Buddhist Teor ofKowledge. London, George Allen and Unwin,
1963.
Jitiri, Hetutattvopaea. Edited by Chattopadhyaya, Durgacarana, Calcutta, University of
Calcutta, 1939.
Joshi, Rsik Vihari , Treatment ofFallac in Indian Logic. Nagpur, Vidarbha Samshodha
MandaI, 1965, pp. 270-301 .
Kajiyama, Yuichi, "On the Theory of Intrinsic Determination of Universal Concomitce
in Buddhist Logic. " Joural ofIndian Buddhist Studies, 7, pp. 354-60.
. "The Avaavi Nirakr1a of PJQit Asoka. " Joural ofIndian Buddhist Studies,
9, no. 1 1961 , pp. 371-76.
KamaIasila, Tattva Sargraha Pij ika, 2 Vois. , Edited by Shastri Dwarikadass, Varanasi,
Bauddha Bharati, 1981 , reprint Delhi, MotilaI Banarsidass, 1986.
Kashyap, Jagdish. (Edited) . Kthivathu. Bihar Pli Publication Board, 1961 .
Katsura, Shoryu, "New Sanskrit Fragments of te Pa1a Samuccaa, " Joural of
Indian Philosoph 3 , March/April 1975, pp. 67-78.
Keith, A. B. "Vasubandhu and His Vooa Vidhi. " Indian Historical Quarterly 4, 1928, pp.
221-27.
Indian Logic and Atomism. New Delhi, Oriental Rprint, 1977.
Kenneth, K. 1. , "Some Basic Misconceptions of Buddhism. " Interational Philosophical
Quarterly 10, no. 1 , March 1969.
Kher, Chitrarekha, "The Concept of Pama1a according to DiInaga and Dharakirti . "
Joural ofthe Oriental Institute 22, no. 3, March 1973, pp. 256-64.
Kitagawa, Hidenoir, (Edited) . "A Reftation ofSolipsism (A Annotated Translation of
the Santanantarasiddhi) . " Joural ofGreater India Societ; 14, no. 1 , 2, 1955, pp.
407-29.
. "A Note on te Methodology in the Study of Indian Logic. " Joural ofIndian
Buddhist Studies 8, no. 1 , 1960, pp. 380-90.
Matilal, B. K. , and Robert D. Evans, Editors, Buddhist Logic and Epistemolog; Studies
in the Buddhist Analysis ofInference and Language. Dordrecht, D. Reidel, 1986.
Misra, Durveka, Dharottara-Padipa. Edited by MaIvania Dalsukabhai, Ptna, K. P.
Jayaswal Institte, 1971 .
Misra, Partasarati , Nia Rtna Mala. Baroda, Gaekwad Oriental Institute, 1937.
Misra, Vacaspati, Niavartiktatarapk. Edited by Dravid Shastri. R. , vols. 1, 2,
Varanasi, Chowkamba Sanskrit Samsthana, 1926.
Mokikaragupt, Bauddha Tark Bhia. Edited by Giri, Raghunatha, Varanasi, Prachya
Prakashana, 1969.
Mookerjee, Satkari, (Edited) . Vois. 1 , 2, Pata, Nav-Nalanda Mahavihara Rsearch, 1937.
Buddhist Philosoph ofUniversal Flux. Delhi, Motilal Banarsidass, 1975.
. Manasa prtaka: A Conundrum in the Buddhist Pama1a System. Dordrecht,
1 32 Bi bl i ography
D. Reidel, 1980, pp. 243-60.
Oliver, J. W. , "Fonal Fallacies and Other Invalid Arguments. " Mind, 76, 1967.
Randle, H. N. Fragments fom Dilnaga. Delhi, Motilal Banarsidass, 1 981 .
Indian Lgic in the Early Schools. Delhi, Oriental Books Rprint, 1 976.
. "A Note on the Indian Sllogism. " Min, 33, 1924, pp. 398-414.
Saradchandra, E. R. , "From Vasubandhu to Santarakita. " Joural of Inian Philosoph
4, no. I , 1 976, pp. 67-77.
Shah, Nagin J . , Aklmk's Criticism ofDharkrti 's Philosoph. Ahmedabad. L. D.
Institute of Indology, 1967.
Shana, Brajanarayana, Bharatfya Darfana main Anumana. Bhopal, M. P. Hindi Granth
Academy, 1 973.
Shastri, D. N. , Te Philosoph of Naa-Vaife#k and Its Confict with the Budist
Dinndga School (Critique of Indian Realism) . New Delhi , Bharatiya Vidya
Prakashana, 1 976.
Shastri Dwarikadas, (Edited) . Suta Nki a. Varanasi, Bauddha Bharati.
. Uddyotakr k Naa Vartk el Adhaana. Kanpur, Bharatiya Prakashana,
1 974.
Stall, J. F, "Correlations between Language and Logic in Indian Thought. " Bulletin of
the School of Oriental and African Studies, 23 , no . I , 1 960 ,
pp. l09-22.
Santarakita, Tatva Samgrha, Edited by Shastri Dwarikdass, Varanasi, Bauddha Bharati,
1 981 , reprint Delhi, Motilal Banarsidass, 1986.
. "The Concept of Pala in lndian Logic" Joural of Indian Philosoph, 2,
no. 2, 1973.
Sugiura, Sadajvio. Hndu Logic as Peserved in China and Japan. University of
Pennslvania Press, 1900.
"Is the Naa Pavesa by DiInaga. " Joural ofthe Ral Asiatic Societ 82,
1 928, pp. 7- 1 3.
"Buddist Logic before DiInaga. " Jourl ofthe Ral Asiatic Societ 81 , 1929,
pp. 451-88.
On Some Aspects of the Doctrines of Maitreanatha and Asanga. Calcutta,
Calcutta Universit, Press, 1 930.
Pe-Dinnaga Buddhist Tets on Logic fom Chinese Sources. Madras, Vesta,
1 981 .
"The Problem of Svalala1a in Sautrantika Epistemology. " Joural of the
Oriental Institute 20, 1 970- 1971 , pp. 216-25.
Th. Stcherbatsk, Buddhist Logic Vois. 1 and 2. Delhi, Motilal Banarsidass, 1993.
Uddyotakara, Nia Vartik. Edited by Thakur, Anant La!, Vaishali, Prakrita Jain
Institute.
Vacaspati, Naa Vartik Tdtpara Tk. Edited by Thakur, Anant L. Vaishali, Prakrita
Jain Institute.
Vadi Deva Suri, Pama1anaatattvaloklankra. Bombay, Jain Shitya Vikas Manda!,
1 967.
Vasubadhu, Abhidha
r
makfa, 4 vols. Eited by Shastri Dwarikadass, Varanasi , Bauddha
Bharati, 1 970-74.
Vatsayana, Naa Bhia. Edited by Thar Anant Lal, Vaishali, Prakrita Jain Insititute.
Vidyabhushana, S. C. , "quisites of a Good Hetu. " Annals ofthe Bhandarlr Oriental
Rsearch Institute, 46, 1965, pp. 43-48.
Bibliography 133
. A Histor ofIndian Lgic. Delhi, Motilal Baarsidass, 1978.
Vinltadeva, NaabiMupk. Edited by Mrinalakanti Gangopadhyaya, Calcutta, Indian
Studies Pst and Pesent, 1971 .
Visvaata, Bhii Pariccheda. Varanasi, Chaukamba Sanskrit Series, 1972.
Wayman, Alex. "The' Rules of Debat according to Aaiga. " Joural ofthe American
Oriental Societ 78, no. 1 , 1958, pp. 29-40.
---. A Reconsideration of Dharmaklrti ' s Deviation from, Diinaga on
Patalibhia. Pne, Annal ofthe Bhanarkr Orental Rsearch Institute, Vo1 . 8,
49. 1977- 1978.
Idex
WORS
Abidhitaviaam, 84.
Abhiva, 77.
Abhilipaamsargaoga pratbhisaprtiti,
30, 31 , 32, 33.
Abhrintam (nonerroneous cognition) ,
29-30.
Adhavasea, 19, 21 .
Agrahiagrihi, 63.
Ajfitirthapraki a, 45.
Anabhilipa, 31 , 44.
Anantarfyak, 64.
Anitati, 3, 4.
Antarendriya, 46.
Anu/oma, 51 .
Anumina, 23, 24, 25, 27, 35, 43f.
Anumea, 53, 70, 71 , 79.
Anumiti jfina, 54, 55.
Anupalabdhi, 54, 68, 69, 70, 99.
Anvaa, 54, 59.
Anvaa jfinam, 50.
Ana, 76.
Anipoha, 8, 49.
Anathinupapatt, 86, 87.
Apadefa, 96.
Apoha, 8, 54.
Arthakriyikritva, 2, 1, 30, 45, 50.
Arthasirupa, 20-21 , 33.
Arthibhisa (arthikira, viayikira,
viaibhia), 17, 18, 43.
Arthidhigati (viaihigat), 18.
Asadhetu, 78, 84.
Asarigdhatvam, 32, 33.
Asapaka or vipaka, 53, 64, 77f.
Asatpratipa!atvam or Aviruddatva, 85.
AsidhaniQga, 99.
Atadviv.rti, 26.
Avaava, 93, 94, 96.
Avaavavaa, 98.
Avaavatraa, 98.
Avlnibhiva, 52, 57, 58, 64, 72-74, 78.
Avisamvidaktvam (nondiscordance),
12, 20f. 32-57.
Avrti, 87, 90.
A
gama, 51 , 52.
A
gamajfina, 44.
A
kina, 52, 65.
A
laavifina, 8.
A
ra-!ilgikmirga, 3.
A
tmasamvedana, 47.
Bihirthavida (Realism) , 7, 14.
Bhrinta jfina, 58, 59.
Bhata, 5.
Bhataalghita, 2.
Causalit, relation of, 69, 75.
Caitta, 47.
136
Change, 1 .
Citta, 5, 47.
Cita salghtta, 2.
Daiivaavi, 93, 96.
Dhamma jfinam, 50
Dhara, 58, 65.
Dhari, 58, 65.
Dissimilarity, 109.
Drva, 23, 28, 66.
Dminta, 94.
Dvairupa, 1 6, 1 8, 21 .
Ekefav.rt, 87, 90.
Ekrupa, 78, 87.
Ek dhikraTya, 53.
Eva, 81 , 84, 85.
Grhitagrihi, 63.
Gmak, 57, 73, 78.
Gma, 57, 73. 78.
Griha, 1 9, 53, 58.
GUTa (Attributes), 23, 28, 66.
Heterolouge. See Asapa!a
Hetu, 51f.
Hetu Caka, 8 , 87.
Hetu vidyi, 49.
Hetibhisa, 34, 52, 55, 78, 84.
Homologue. See Sapa!a
Identity, relation of, 69, 75.
Indriya, 15, 45.
Indriya prta!a, 43-46.
Indriyirtha sanTikara (rej ection by
Dimaga), 27.
Inference (anumina), 26f.
Jiti, 23, 28, 35, 66.
Jifisi, 67, 93, 94, 96.
Jfina, 18, 34.
Jfinitmak, 65.
Jfipak, 11 .
Kalpani, (See also viklpa,
siminala!aTa), 23, 26-28, 30, 33,
43, 45, 58.
Kalpanipotha, 27f.
Kra, 23, 28, 66.
KrTa, 14, 22.
Kirk, 1, 12.
KiraTa, 1 3, 69.
KiraTa simagri, 14.
Kirmic-force, 3.
Index
Kowledge, (samakjfina) Theory of, ix,
lf. , 32, 54, 55.
1pes of, 12, 23.
Kira, 68, 69, 99.
KaTa, 2.
KaTa-santina, 3.
KaTikvida, 4.
LiTga, 53f. , 57, 64, 65, 68, 69. 70, 71 ,
77, 78f.
LiTgin, 57f. , 67, 69, 70, 71 , 78.
Madhamaprtipad (middle pat) , 6.
Manas, 1 5.
Manovifina, 44, 46, 50, 51 .
Manovifina dhttu, 46.
Minameodaa, 4.
Minasa prata!a, 43-46.
Minidhinimeasiddhi/, 22, 24
Meidhiniminasiddhi/, 8, 22, 24.
Mithi, 1 .
Modes, 1 .
Momentariness, 15, 1 6.
Nima, 23, 35.
Nima skndha, 2.
Nintariyak, 64.
Ngamana, 51 , 93f.
Nggaha, 51 .
Nrikrjfinavida, 1 9.
Nirupidhik, 86.
NrviTa, 3, 4.
Niriklpak, 29, 35.
Nicitam, 81 .
Noble Trth, 2.
Pa!a, 53, 63f.
Pa!adharmati, 65-68, 71 , 7, 84, 86, 87,
96, 106.
Pa!ibhtsa, 106.
PafcakrTi, 54.
Index
Pancarapa, 78, 84, 85.
Pancaskndha, 3.
Pncavaavi'vdka, 51 , 52, 94, 95.
Pari rtha sat, ix. 7.
Pratantra, 7, 8.
Pramaria, 94.
Prarthtnulina, 9, 49, 51 , 52, 59, 63,
64, 98.
Parklpita, 7, 8.
Parinipanna, 7, 8.
Prolajnna, 26, 30, 49, 65.
Pa!kmma, 51 .
Paralirhik, 54.
P, 7, 48, 51 , 52, 9Of.
Paki ana, 52, 65.
PaliTphala, 1 3f. , 55, 56.
PaQlata, 1, 49.
PaQavta, 38, 43.
Pali Qavavastht, 8, 24, 60, 62, 85,
PamaQabhaa, 56.
Pamea, 50, 52, 56, 6.
Patibhaa, 16, 19-20, 27, 29-31 , 61 .
-Aniyata, 19.
-Nyata, 1 9.
-Dhanaklrti's analysis, 20.
-DiIniga's analysis, 20.
Patibhaaprtlti, 27, 30-31, 35.
Patitaamutpta, 4, 6.
Pativiaa, 29.
Patala, 8, 23f. 49.
-Dhanalrti' s defnition of,
28f.
-Diniga's defnition of, 27.
-Vasubandhu's defnition of, 26, 29.
-objects of, 30.
-tpes of, 43f.
Pataldbhtsa, 34.
Pataa, 15, 26.
-Ahipati, 1 5, 26.
-
A
lambana, 1 5, 19, 26.
-Samanantar, 1 5, 26.
Paojana, 93, 94.
Probandum. See Stha
PamaTya, 20, 54.
PdmdTyavta, 43.
Parvavat anumana, 69.
Rality
-theory of, 12.
-tree levels of, 7, 8.
Rapa, 1 8.
Rapakndha, 2.
Sahetu, 68, 78, 84.
Sahabhahetu, 5, 14, 15.
Sanatara prtaa, 45.
Samthi, 6, 48.
Sa sknha, 2.
SaTkr skndha, 2.
Samvrt i sat, 6.
Saskra, 3.
Saraa, 93, 94, 96.
Saraavudaa, 93, 94.
Saneha, 67
.Santana,. 3 .
Sanghtta, 2.
Sapala, 53, 64, 76f.
Sapalesattvam, 84, 86, 87.
Saragatd, 48, 53.
Saviklpak, 32.
Sthana, 7, 68, 98, 105.
Sthana, 105,
SthantTga, 99.
137
Stha, 53, 64, 65, 68, 70, 71 , 72, 74.
Sakr, 19.
Sakrajianavta, 7, 20, 21 , 22, 56.
Samartha, 61 .
SdmanallaQa, 8, 16, 20-21 , 24, 34, 53,
58, 60, 61 .
Sarv.rik, 54.
-arthasdrapa, 22.
-as prmaQa 14.
-viklpa, klpana, 25, 26.
-in early Buddhism phlosophy, 14.
-justifcation for its acceptance, 21 .
Sdrapa, 5, 6, 7, 14f. , 32, 54.
Sithaia, 67.
Sopthik, 86.
Svabhiva, 6.
Svabhiva hetu, 68, 69.
Svabhivaprtibandha, 73.
Svabhivanumana, 53, 66, 98.
Svakrta, 55.
Svatovavartana, 8.
Svarthanuman 9, 50, 52, 53, 63, 64, 97.
138
Self, Buddhist conception of, 3 .
Substance, 1 .
. Sufering, 4.
Svala/all, ix, x, 8, 16, 20, 24f. , 53,
58, 60, 61 , 76.
Svapraki a, (svaampraki a,
svasamvedana) 14, 16, 43, 47.
Svasamvedana prta/a, 43, 46.
Svaalvitti, 14f.
Svaata/prtmilya, 1 6.
Svata/prtmtlyavtda, 55, 56.
Svtbhta, 16f. , 43.
SVtlra, 1 7f.
Sabda, 52, 60, 94.
Sabdttmak, 65.
Sakaprtpti, 93, 94.
Seavat anumtna, 69.
Sila, 7, 48.
Srmall tradition, 3 .
Suna, 4, 6.
-prapaica sana, 6.
-svabhtva sana, 6.
Tautpatti, 66, 74, 76.
Tanht (longing), 2.
Tark, 75, 94.
Tattulya, 79.
Tattva-jnna, 11 , 1 8.
Tttta, 74, 75.
Trirapa, 51 , 52, 53, 64, 65 , 77-79,
85-87.
Tratvaavi, 52.
Trt, criterion of, 14.
Udtharall, 51 , 84, 9Of.
Upamina, 52, 60, 94.
Upanaa, 51 , 84, 93, 94, 96.
Vaidharmell , 105.
Vairapa, 21 .
Vtda-vidhi (art of debate), 49.
Vtda sttra, 49.
Vedantskndha, 2.
Viitna, 6, 7, 1 5, 56.
-triple division of, 19.
Viiapti-mitra, 19, 20.
Viklpa, (See also stmtnala/ala), 25f.
Vipa/a, 77.
Vipa/tvtvrt , 84 .
Vipa/e asattvam, 86, 87 .
Vruddha, 77.
Viaa, 15, 1 8.
Vaajitna, 1 7, 1 8.
Viaajiina-iina, 1 7, 1 8.
Vatbhta. See Arthtbhtsa.
Vatdhigama, 14.
Vatlrtt, 55.
Vyatireka, 54, 59, 94, 106.
Vavatttak, 32.
Vtpak, 74, 87, 90.
Index
Vtpti, 53, 55, 57, 58, 65, 71 , 72, 74, 75,
96.
Vtpa, 68, 74.
Ygi prata/a, 43, 44, 48.
Yogi jitna, 44, 48.
AUHORS, WORS, AN
SCHOOLS
Abhidhamma, 51 .
Abhidhamma literture, 46. 50.
Abhidharma Ksa, 7.
Abhisamatlallr Ktril, 7.
Avaita Vedtnta, 2.
Aklal, 85, 86.
Anantavira, 86.
Anttmavtda (no-substance ontology), 2.
Anumina Satta, 50.
Arca!a, 62, 74.
Asalga, 7, 8, 33, 51 , 52.
Aviddhakrll , 97.
Alambana Pari/t, 9.
Atamavtda (substance ontology) , 2.
Bhartrhari, 54, 59.
Bodhisattvacart Nirdesa, 7.
Ctrvtk, 3 , 49, 54.
Codant Pakrala, 9.
Chi, 82, 83, 84.
Dharakrti, 9, 1, 12.
-on inference, 49f.
-on knowledge, 1, 12, 14.
-on perception, 26
Index
-on prmtla a pramtlaphala 14.
-on pratibhtsa, 20.
-on strupa, 14.
on types of perception, 46f.
Dharmottara, 12, 14, 20, 30, 31 , 33, 34,
, , , , , M, , n, m.
-on prmtla, 14.
-on prmtla and prmtlaphala,
13, 14.
-on process of apprehension of an
object, 12.
-on strapa, 14.
-on types of perception, 46.
Dilntga: ix. 7, 8, 9.
-on inference, 49ff.
-on kowledge, 13f.
-on perception, 26f.
-on prtibhtsa, 20.
-on strapa, 14, 2l .
-on tpes of perception, 43.
Durek, 46.
Gngeia, 8, 55 .
Gutam, the Autor of Nia Satra, 29.
Hattori, 17-19.
Hemacandr, 86.
Hetu Bindu, 9.
Hetu Cakra Damaru, 9, 76, 81 , 87.
Hfnatna, 4.
Jainism, 2, 78, 85, 86, 98.
Jaanta, 95, 96, 97.
Jitnairf Mitr, 54.
Kmalaifla, 32, 34, 35, 54.
Kthtvatthu, 51.
Kit, 52.
Mahttna, 4.
Mahttnbhiharaukaangititst,
5l .
Maitrea, 7, 8.
Materialism, 4,
Mtdhavtctra, 5.
Mtdhamik, 4, 6.
Mtnameodaa, 4.
MfmtTst, 2, 26, 60, 94.
Motkrgupta, 54, 70, 99.
Mookherjee S. , 98.
Naiytik, 60, 64.
Nava-Nta, 8, 67.
Ntgtrjuna, 6, 51 .
NtrtaT Bhatt, 4.
1 39
Nta, 51f. 67, 69, 70, 84, 85, 95.
Nta Bindu, 9, 11 , 20, 30, 44, 48, 53,
63, 64, 70, 98.
Nta Bindu Tk Tippalf (Durek Miira)
46, 50, 51 .
Nta Muka, 9, 33.
Nta Paveia, i, x, 9.
Nta Ststr, 49.
Nta Satra (Gutam) , 29, 98.
Nta-Vaiieiik, 2.
Nta Vtrtik 'k, 68.
Nikas, 50.
Pariiuddhi, 55.
Pttrsvtmi, 86, 87.
Pabhtcandr, 86.
Pajitkragupta, 54
PakrT, 5l .
Pakraltra Vtcti tstra, 51 .
Pamtla Samuccaa, 9, 47, 55, 60, 73,
lO.
PT Vtrtik, 9, 12, 44, 48, 53, 70,
98.
Pamtla Viniicaa, 9.
Rtnakr, 54.
Sambandha Parft, 9.
Santtntntara Siddhil, 9.
Saptaaiabhamiitstr, 7, 51 .
Sarvtstivtda, 14.
Sauttntrik, 4, 5, 6, 9, 14, 15, 16.
Stkraj ianavtda, 6.
StTka, 2.
Stntarita, 14, 33, 34, 35, 44, 54, 86,
87, 96, 97, 99.
Stcherbatsky, T. , 14, 16, 33, 45, 52, 64,
82.
Suta Nka, 50.
Satrtnta, 6, 7.
140
STdhar, 96, 97.
Sunavdda, 6.
Tark Sctra, 7, 51 , 52.
Tatva Samgraha, 86.
Tipi!ak, 3, 5, 50.
Tucci, 51 .
Udaana, 55.
Uddotakra, 52, 79, 80, 96, 97.
Upaniiadic thought, 1 .
Vaibhik, 4, 5, 1 5.
Vaubandhu, 7-9, 26f. , 34, 43, 49, 51 ,
52, 64, 79.
Vacapati, 55, 95, 97.
Vdda Vdhi, x, 7, 51 , 6.
Vdda Kiaia, 7.
Vdda Marga, 7.
Vddideauri, 86, 93.
Vedantn, 54.
Vbha, 5.
Vidyabhusana, 52.
Vdyananda, 86.
Viinavdda, 4, 7, 14, 21 , 43, 56,
Vnfadea, 33.
Ygacarabhumiicta, 51 .
Index
Ygacara (jinavt), 4, 6, 9, 23, 47.
Abut te Ator
S. R. BHT is Senior Professor of Philosophy at Delhi University. His reseach
interests include Indian philosophy, logic and epistemology, social and political
thought, and philosophy of religion. He has published numerous scholaly acles
and seven books, including Te Philosophy of Pancartr (1968) and Knowledge,
Values and Education ( 1986).
AU MHROJ is senior schola of Philosophy at Delhi Universit. Her research
interests include Indian Philosophy, logic and epistemology, philosophy of reli
gion, and social and political thought. She has published scholarly aticles on
Nehru and Buddhist Philosophy.