topics: |  philosophy | induction | india

Excerpts

the problem of induction is to try to justify the claim about our
knowledge of unobserved cases from our knowledge of observed cases.

in other words, the question is: can our experience of past and present
particular instances make our generalized claims about all instances
including past, present and future unobserved instances reasonable,
reliable and acceptable? for example, when we observe in some cases that
smoke is produced by fire and never observe a case where smoke is
produced without fire, we may generalize that wherever there is smoke,
there is fire.  such induction includes a claim about all smokes — past,
present and future — that they are produced by fire though only a limited
number of actual cases have been and can possibly be observed. do we then
have the right to claim that smoke is always produced by fire? in other
words, can our observation of co-presence of smoke and fire in some cases
make it reasonable, reliable and acceptable that smoke never exists
without fire?

chArvAka's lost argument

in indian philosophy the problem arose in the context of examining the
status of a kind of inference as a source of knowledge. the view that a
paradigmatic kind of inference is not a source of knowledge (and by
extension that no kind of indirect awareness is a source of knowledge)
was forcefully presented by philosophers of the cArvAka school, many of
whom held that perception or observation of particulars is the only
source of knowledge.  unfortunately, however, the writings of cArvAka
(sixth? century BCE) and his principal followers are lost (except for
jayarasi, to whom we turn later). but, fortunately, the cArvAka viewpoint
has been preserved by their philosophical opponents, including the nyAya
philosophers.  [it is common in the sanskrit philosophical tradition to
state rival views clearly and precisely. the rival view is called
pūrva-pakSa (the predecessor's view) and sometimes even contains
improvements on the original. the favored view is uttara-pakSa (the
successor's view).

we look at the great nyāya philosopher udayana for an account of the
cArvAka position. like other sanskrit philosophers, udayana (eleventh
century) writes in a compact style; hence some explaining has become
necessary.

   cArvAka says: that which cannot be perceived does not exist. the opposite
   exists.  god, etc., are not so; therefore, it should better be held that
   these do not exist. it may be objected that inference, etc., will then be
   eliminated. but this is not unwelcome.

   objection: but then common activities would be impossible.
   reply: no. that can be carried out on the basis of EXPECTATION
   (sambhāvanā) alone. coherence is mistakenly thought to justify the
   claim of knowledge. (NK 334)

	[NK = Udayana, Nyāyakusumāñjali, with Nyāyabodhanī, Prakāśa, Prakāśikā
	      and Makaranda, Chowkhamba, Benares, 1935]

in the cArvAka view, if something cannot be perceived by anyone at any
time whatsoever, then, since perception is the only source of knowledge,
it cannot be admitted to exist. since god and so forth are imperceptible,
it is better not to admit that they exist. only what is perceived exists
(not that all that is perceived exists). since it is unnecessary to admit
existence of anything imperceptible, it is also unnecessary to accept
inference (or any other indirect means) as a source of knowledge.

can one act if one does not make inferences?

is not inference indispensable even for common activities, such as
searching for fire after seeing smoke?  the reply is: no. it is indeed
necessary to go beyond what is perceived at a given time and form
opinions about the past as well as expectations about the future. all
such activities can be fully explained in terms of such expectations.

for example, one searches for unperceived fire after seeing smoke based
on expectation that there is fire.

it is both unnecessary and unjustified to claim that there is inferred
knowledge of fire in such cases.
when fire is actually found, does not that justify, because of the
coherence (samvāda) between what was previously expected and what is now
perceived, that there is knowledge of fire, so that acceptance of
inference as a source of knowledge is necessary?

the reply is: no.  success of action prompted by expectation does not turn
expectation into knowledge. but such success and coherence suffice to
generate confidence in expectations and make them appear as
knowledge.  "appearing as knowledge" is all that is needed to account for
such activities.

[AM: takes a very strong view of "knowledge" - as indubitably true.
   "appearance of knowledge" may approach knowledge that is highly
   probabile. ]

Expectation vs Knowledge

rucidatta, who wrote the prakAsha commentary on the nyAyakusumA~njali,
has described expectation as a doubt one side (koti) of which is
stronger (utkata) than others (NK 334). if each side of expectation is
equally matched, expectation would not lead to any action. but if one
side is stronger than the others, expectation may lead to action. for
example, when one sees smoke, one does not have any rational grounds for
being sure that there is fire, but may nevertheless have a strong
expectation that there is fire. this is a doubt with two sides, viz.,
that (1) there is fire and that (2) fire is not there. but the two sides
are not equally matched; the first is stronger than the second, for fire
has been observed together with smoke on many occasions. hence it may
very well lead to action of procuring fire.

the cArvAka philosopher argues further:

    since there is no discriminating factor, how can it be known that
    although there is deviation in a certain case, there is no deviation
    in some other case? thus, since there is no reason that can settle
    the matter one way or the other, the observation of togetherness
    itself is the ground of apprehension of deviation (vyabhicāra). how
    then can it be groundless? it may be said that there is deviation in
    some cases and not in some other cases due to the nature of things
    and that it is the nature of things which provides the discriminating
    factor. but by what signs can the nature of things be determined with
    certainty? this question should be considered carefully. for what is
    confirmed in hundreds of cases is also found to be refuted. it may be
    said that where no counterexample is known, there that is so [i.e.,
    one has a proper reason for generalizing]. but from the fact that no
    counterexample has been found so far, who can legislate that none
    will be found anywhere at any time? [quoting udayana (NK 339)]

several arguments are compressed in this passage. the nyāya philosophers
have accepted the observation of co-presence (sahacāra-darśana) as a
method of generalization. it is pointed out first that the method cannot
give any valid reason for making such a claim. even when two things have
been observed together in some cases, the one that is supposed to be
pervaded is sometimes found to exist without the other (the supposed
pervader).  this establishes the fact of deviation and falsifies the
general claim. hence one cannot have any reason that this is not so in
other cases when two things are observed together, for there is no
objective ground for discriminating between the two situations, viz.,

  (1) two things are together sometimes and separated sometimes, or
  (2) two things are together always.

accordingly, no generalization based on observation of co-presence can be
justified. but then since there is no ground for generalizing, no such
inferences can be sources of knowledge, for they all require at least one
general premise that the probans is pervaded by the probandum.  since the
premise is baseless, the inference is baseless too.


the cArvAka goes on to say: "deviation and nondeviation follow
respectively from presence and absence of adjuncts (upādhi); but the
determination of that [i.e., determination of the absence of adjuncts] is
impossible" (NK 339).

to explain: co-presence of two things or characteristics may depend on
availability of adjuncts or additional third factors; if so, at least one
of those two things/characteristics will be found without the other when
the third factors are missing. for example, if one has observed every
earthen vessel to be brittle and generalizes thereby that all earthen
vessels are brittle, one overlooks that brittleness is not due to being
earthen or being a vessel, but due to other factors, such as being built
or baked in certain ways. in absence of those other factors, an earthen
vessel will deviate from brittleness (i.e., an earthen vessel will not be
brittle), and the generalization will be falsified. however, if
co-presence of two things or characteristics is not dependent on any
third factor, the nyāya holds, they are nondeviant and the generalization
that one of them is pervaded by the other is true. thus one must
carefully observe if any third factors are involved and "elimination of
adjuncts" (upādhi-nirāsa) is a requisite step for generalizing. the
cArvAka argues that while some third factors may be detected and
eliminated, one cannot be sure that all third factors are eliminated. so,
no empirical generalization is justified.

while an ADJUNCT is anything that leads to deviation of the MARK from the
PROBANDUM, in the narrower, technical sense, it is defined as "that which
pervades the probandum but does not pervade the mark" (NK 352). This
definition may be explained with the help of the following stock
example. while it is true that wherever there is smoke there is fire, it is
not true that wherever there is fire there is smoke. this is because fire
emits smoke only if the fuel is wet. thus wet fuel (_ārdrendhana_) is the
third factor on which co-presence of fire with smoke depends. the detection
of the adjunct vitiates the generalization and also the inference of smoke
from fire.

PROBANDUM =  smoke [evidence, premise]
	  <-- does not agree w L.  'probandum' (that which is to be proved)
MARK = fire [conclusion]

ADJUNCT pervades the probandum (wherever there is smoke there is wet fuel),
   but the adjunct does not pervade the mark (fire may be found without wet
   fuel, as in an electric heater).

from blurb

"Kisor Kumar Chakrabarti demonstrates convincingly that Indian
philosophical analyses of induction out-distance and often chronologically
anticipate Western treatments. Furthermore, he shows that the problem of
the justification of induction is more satisfactorily solved in Indian
thought than in contemporary analytic philosophy."
			 --Karl H. Potter, University of Washington.

author bio

kisor kumar chakrabarti is professor of philosophy and chair of the
department of religion and philosophy and distinguished scholar in
residence of the davis and elkins college (elkins, w.va).  he has taught at
university of calcutta, university of maine, kutztown university, ferrum
college and university of california at berkeley. he has held faculty
research fellowships at the australian national university, the indian
institute of advanced study, shimla, the university of pittsburgh and the
institute for advanced study, princeton. he is the author of the logic of
gotama, definition and induction, classical indian philosophy of mind and
introduction to buddhism and hinduism. he has contributed numerous articles
to the encyclopedia of indian philosophies, the encyclopedia of hinduism
and the encyclopedia of religion and published more than fifty papers in
philosophical journals and anthologies. he is editor of the journal of
indian philosophy and religion.