Académique Documents
Professionnel Documents
Culture Documents
Personal Details
Principal Investigator Prof. A. Raghuramaraju Department of Philosophy,
University of Hyderabad, Hyderabad
Paper Coordinator (1) Ranjan Mukhopadhyay (1) Visva-Bharati, Shantiniketan
Description of Module
Module ID 7.33
Prerequisites Nyya theory of Anumna
Objectives To understand the defects of bad inferences from the Indian philosophical
perspective
Key words hetvbhsa, vyabhicra, bdha, svarpsiddhi, paksiddhi, satpratipaka
2
1. Basic Notions
As an epistemic process, inference is extremely useful and is a great source of valid cognitions
(prams). But all inferences are not correct. An inferential defect (hetvbhsa) is the factor that is
responsible for invalidating or corrupting an inference.
Before we discuss the inferential defects, we need to know a few other technical terms of Indian
epistemology. Like other intentional entities (saviayaka-padrthas), a cognition or jna is about
something. That something is called the content (viaya) of the cognition. When I cognize a pot, the
content of my cognition is that very pot. In other words, my cognition is about that pot (ghaa-viayaka).
The content of a cognition need not always exist in the real world. The content of an illusory cognition
(bhrama or a-yathrtha-jna) is not a real object. I may mistake a rope for a snake. That illusory snake
does not really exist.
Another relevant notion in this context is the notion of rivalry (virodha). Sometimes a cognition
C1 rivals another one, say C2, by blocking C2. Suppose, after having observed many cases of co-
existence of smoke and fire, and a few red-hot iron-balls, I have concluded that fire can exist without
1
For the sake of convenience, I shall use cognition as a count noun. Thus a cognition will henceforth mean a
piece of cognition.
2
y pervades x if and only if any case of x is a case of y.
3
When somebody infers y from x, y and x become the Target (sdhya) and Reason (hetu) respectively. The locus of
inference is called the Site (paka). When after seeing smoke on a hill, somebody infers that the hill has fire, fire,
smoke and the hill become the Target, Reason and Site of their inference respectively.
3
being accompanied by smoke, although smoke must be accompanied by fire.4With this conclusion in
mind, I cannot infer smoke from fire. In this case, the cognition, smoke does not pervade fire blocks the
inference of fire from smoke. Thus the former becomes the rival (virodhi) of the latter. We can say that a
cognition cannot be generated in the presence of its rival cognition (virodhi-jna).
For the Old school Naiyyika (prcna-naiyyika), a hetvbhsa is a defective Reason (dua-hetu). But
according to the Neo Naiyyika, a hetvbhsa is the defect of an inferential Reason (hetu-doa). In this
essay, we are going to follow the Neo School.
The definition is the following. The property of being a hetvbhsa [of the inference I] is the
property of being the content c, by which a cognition, that possesses c, rivals I. Example: contents
such as deviation (vyabhicra); for a cognition having deviation as its content rivals an inference.
Here content should be understood as qualified content (viia-viaya). Now, even if an
illusory cognition, having bdha as its content, rivals an inference, there would be no problems.5
Vivantha uses a few technical terms in the passage quoted above. Let us try to understand some of
those. According to Nyya, a snake is not a simple object; it is rather an object-complex. It is a property-
holder (dharmin) that has the property called snake-ness (sarpatva). Snake-ness resides in the snake
through the relation of inherence (samavya). Thus every object can be analyzed in terms of three
aspects; the property, property-holder and relation. When the snake gets perceived, the object-trio appears
in the content of the perception. The property, property-holder and relation appear in the perception as the
4
A red-hot iron-ball (tapta-aya-golaka) has fire, although it does not emit smoke. This means that
smoke does not pervade fire, since all the cases of fire are not cases of smoke.
5
lakaantuyad-viayakatvenajnasyaanumiti-virodhitvatattvam. tath hi ---- vyabhicrdi-
viayakatvenajnasyaanumiti-virodhitvttedo. yad-viayakatva ca yda-viia-
viayakatvabodhyam, tenabdha-bhramasyaanumiti-virodhitve pi nakati.Nyya-siddhnta-
muktval, The Chapter on Inference. See str (1968, pp. 375-379).
4
qualifier (vieaa or prakra), qualificandum (vieya) and cognitive relation (sasarga) respectively.
In the case of a valid cognition (pram) having the form, this is a snake, the qualifier (which is snake-
ness) is related to the qualificandum (which is just the yonder object denoted by the word that) in
reality. But in the case of an illusory cognition having the same form, something else happens. Due to
insufficient light or physical defects, the cognizing subject sees something, maybe a rope, without
capturing its essential property. But the resultant cognition must have a qualifier, although the qualifier
has not been captured by the sense organs of the subject. To fill in the blanks in the cognition, the
memory of the subject provides the cognition with a qualifier through an extraordinary connection called
jna-laka. Roughly speaking, the subject remembers a property they have known before. That
property is snake-ness in this case and the subject cognizes an illusory snake. The point to be noted here is
this. When the cognition is valid, the property that appears as the qualifier (i.e., snake-ness) really
qualifies the property that appears as the qualificandum (i.e., the object denoted by the word this), since
in reality the cognized object is a snake. On the contrary, when a rope is mistaken for a snake, i.e., when
the cognition is illusory, the qualifier snake-ness does not really qualify the qualificandum, which is a
rope. In reality, snake-ness is not related to a rope through inherence. The gist of the whole discussion is
this: when the content of the cognition C is qualified (viia), Cis valid or correct; the content of an
illusory cognition cannot be qualified, since its qualifier does not qualify its qualificandum in reality.
What is the relevance of a qualified content in our discussion? The Neo Naiyyika wants to say
that even an illusory cognition may rival an inference I by blocking the genesis of I. That being the case,
the content of that illusory cognition would not be considered an inferential defect of I. Suppose, Ram
infers fire from smoke on a hill. But for some reason, John cannot see any fire there. That means, John
has an illusion of the absence of fire. That being the case, the content of Johns illusory cognition having
the form, there is no fire on the hill would not be an inferential defect of Rams inference, this hill has
fire, since it has smoke. When the blocking cognition is valid, then and only then does its content
become a defect of the blocked inference. I reiterate: only a valid cognition possesses a qualified content.
A cognition having non-qualified content may block an inference; but the content of that cognition will
not be an inferential defect.
Let us consider a genuine inferential defect. Any inference is based on the cognitions of pervasion
and Site-located-ness. Thus the inference, z has y, since z has x, is based on the following cognitions:
(1) Pervasion: Any case of x is a case of y.
(2): Site-located-ness: z has x.
One cannot inferentially claim/assert z has y, since z has x without asserting/claiming (1) and (2).
Suppose somebody infers smoke from fire. The moment they inferentially assert that this hill has
smoke, since it has fire, they also assert:
5
6
According to Bhpariccheda, savyabicra is of three typessdhraa, asdhraa and
anupasahrin, and asiddhi too is of three types--raysiddhi, svarpsiddhi and vypyatvsiddhi.
6
Bdha: This is nothing but the Site without the Target.7 Suppose somebody infers fire at a place where
there is no fire. In this case, the Target itself is absent from the Site. This defect opposes the inference
itself. i.e., when the inference is z has y, and there is no y in z, the defect of the inference would directly
oppose the inference itself, not its pervasion or Site-located-ness.
Svarpsiddhi: Suppose somebody infers, z has y, since z has x when z does not really have the Reason
x. Here the absence of x in z is the inferential defect called svarpsiddhi. Example: The defect of the
inference, This lake is a substance, since it has smoke is the absence of smoke in the lake.8
Paksiddhi: An un-established (aprasiddha) Site, i.e., a Site that is not found anywhere is the defect
called paksiddhi. The defect of the inference, A hill made up of gems has fire is the hill made up of
gems (since such a thing is not found anywhere).9
In order to accommodate the cases in which two equally strong pervasions oppose each other, the
definition of the inferential defect should be modified a bit. The modified definition is the following:
7
sdhya-nyoyatrapakastvasaubdhaudhta. Bhpariccheda by Vivantha, The Chapter on Inference. See
str (1986, p. 406).
8
hradodravyadhma-vattvdatrsiddhirathpar. See str (1986, p. 406).
9
paksiddhiryatrapakobhavenmai-mayo giri. Ibid, p. 404.
7
An inferential defect of the inference I is the content (viaya) (i) that belongs to the cognition C,
which rivals I, and (ii) the invalidity of Chas not been established (agrhita-aprmnyaka). A piece of
cognition whose invalidity has not been established may be called an un-invalidated cognition. Such a
cognition is either a valid one, or something, which has not been invalidated so far.
The vyabhicra defect of the inference, this hill has smoke, since it has fire is the fact that
smoke does not pervade fire. This fact meets both the criteria, since the cognition that represents it rivals
the inference and the cognition is a valid one. The invalidity of a valid cognition cannot be established,
since it does not exist. Thus the cognition is un-invalidated. And the rivaling pervasion-cognitions in our
Coccinistercus case are: anybody who eats strawberries everyday suffers from a disease called
Coccinistercus and anybody who eats blackberries everyday never suffers from that disease. Neither of
these cognitions is known to be invalid. Therefore, both are un-invalidated. Thus the modified definition
accommodates the cases of sat-pratipaka.
According to some scholars, sat-pratipaka is the content that belongs to the cognition C1, which
is such that (i) C1s invalidity has not been known, (ii) C1 is about something that is pervaded by
the absence of the Target, and (iii) C1 is co-temporal with another cognition C2, which is about
something that is pervaded by the Target and C2s invalidity also is not yet known.10
Suppose Ram has attained the cognition C2 that has the form, z has y that is pervaded by x. (y is
pervaded by x means any case of y is a case of x.) Ram does not know whether C2 is invalid. At the
same time suppose he also attains the cognition C1 that has the form, z has w that is pervaded by
the absence of x. Ram does not know whether C1 is invalid. In this case, C1 and C2 would mutually block
each other and there would be no inference at all, since y and w would try to establish x and the absence
of x respectively at the same time. Here the content of C1 is the sat-pratipaka defect of the inference, z
has x, since z has y, and likewise the content of C2 is the sat-pratipaka defect of the inference, z has the
absence of x, since z has w.
5. Conclusion
10
aghtprmyaka-sdhya-vypya-vattvopasthiti-klnghtprmyaka-tad-abhva-vypyavattvopasthiti-
viayas tathetyanye. Nyyasiddhnta-muktval. See str (1968, p. 390).
8
We may visualize the classical Indian logical system as a large cylindrical machine with an inference-
generator and a few filters. At one end of the machine, the generator is located. It has a schema, which is,
z has y, since z has x. One can instantiate this schema by replacing x, y and z by any entities. Thus one
may get, This hill has fire, since it has smoke or This hill has smoke, since it has fire or anything else.
Each instantiation is an inference. There are a number of filters in the cylindrical system, and those are
associated with a description of the world. The description or database consists of several statements.
Suppose the following statements are part of the database of a logical system:
S5: It is not the case that every case of fire is a case of smoke.
We would assume that none of the above statements is known to be false (i.e., we do not have evidence
against the validity of these statements). Of course all of them need not be valid. Now from S4 and S3, we
may infer Hill3 has fire, Hill4 has fire etc. Those inferences would be valid, since no filter would block
them inside the cylinder.
Suppose Ram tries to infer, Hill1 has fire. The Site-located-ness-cognition associated with this
inference is, Hill1 has smoke, which is blocked by S1. The inferential defect of this inference is the
content of the cognition, Hill1 does not have smoke. Let us also assume that the content is a fact with
respect to the world described in the database. The name of the defect is svarpsiddhi.
In the cylindrical system, there is a particular filter that would be activated if an inference that has
been generated has a svarpsiddhi defect. Thus each defect is connected with a filter. If an inference is
defect-free, no filter in the cylinder would be activated and the inference would freely go out of the
system and get validated. But a defective inference cannot pass all the filters of the system and would not
9
get validated.
The western logical systems are formal, whereas the Indian system is associated with a database/
world. The defect of an Indian inference is not just a structural/ formal defect; it is either a fact or a state
of affairs that seems to be right. The effort to understand the Indian logical system in terms of the western
logical terms may be a little misguiding. In this essay, we tried to see the idea of logical defects from an
Indian perspective.