tasting a single grain). It is not formally valid, but it is a pervasive and powerful
species of informal reasoning. We will see in the next section that it acquires
particular significance in the context of a debate.
There is also an intriguing example from Va ̄tsya ̄yana I have not yet men-
tioned. This is the inference of the “residual” by elimination.^33 Wondering
whether sound is a substance, a quality, or an event, and finding reasons to deny
that it is a substance or an event, one draws the conclusion that it is a quality.
What is interesting is that the early writers are well aware that not all warranted
inference can be reduced to the causal model. An adequate theory of inference
has to find a description of the extrapolation-warranting relation at a level more
general than that of the causal.
The difficulty in finding adequately general bases for extrapolative inference
encouraged a skepticism about their existence. In one form, scepticism about the
possibility of rational extrapolation is just the claim that there is no adequate
extrapolative basis. It is the view allegedly of the Loka ̄yata “materialist,” and
most notably of the sceptic Jayara ̄s ́i (ca. ad600). An early version of the skeptic’s
argument is recorded in the Nya ̄yastu ̄tra (NS 2.1.37):
[Objection:] Inference is not a means of knowing, because there may be errancy
arising from embankment, damage and similarity.Gautama is referring to the examples of the swollen river, the ants, and the
peacock. His point is that the observed facts admit of different explanations: the
swollen river might have been caused by a dam or embankment further down-
stream, the ants might be carrying their eggs because their nest has been
damaged, and the peacock’s cry might have been made by a human or animal
impersonator. There are alternative causal explanations for each perceived
event. This is a standard skeptical move. The sceptic introduces an alternative
possible explanation (that our experiences are all dreams or produced by an evil
genius, that the world was created five minutes ago with all its fossil records) and
then claims that the existence of such an alternative explanation shows that one
is not entitled to assume that the common-sense expanation is the correct one.
Gautama’s reply is interesting (NS 2.1.38):
No. Because it is a different thing from a [mere] partial case, fear, and [mere]
similarity.The compact response is that there is an observable difference between a river
which is swollen because of upstream rain and one which is swollen because of
a downstream blockage, a difference between the orderly procession of ants
when it is about to rain and their fearful scurrying when the nest is disturbed,
and a difference between a genuine peacock’s cry and that of an impersonator.
The implication of Gautama’s reply for the rationality of extrapolation seems
to be this. An extrapolation from the seen thing to the unseen thing is rational
when there is a relation of some appropriate (as yet unspecified) sort between
430 jonardon ganeri