ais likeR 1 ba is likeR 2 c
Tb ~Tc
\Ta \~TaAs an argument, the rejoinder seems to follow the very same pattern as the orig-
inal one, so why is it false? The existence of such rejoinders shows that mere like-
ness is not sufficient for good argument. The likeness has to be of the right type.
When is the likeness of the right type? The Nya ̄yasu ̄tra’s very cryptic comment
is that the “right type” is the type displayed by the relationship between a cow
and its genus cowhood. Va ̄tsya ̄yana, the commentator, is unclear and confused
on this point. He does, however, make one important observation:^39
If one proceeds to establish the required inferable property on the basis simply of
likeness or unlikeness then there will be lack of regularity (vyavastha ̄). Irregular-
ity does not arise with respect to some special property. For something is a cow
because of its likeness with another cow, which likeness is actually cowhood, not
the cow’s having dewlap, etc. It is because of cowhood that a cow is unlike a horse,
etc., not because of a difference of particular qualities. This has been explained in
the section on the limbs of a demonstration. In a demonstration, each limb serves
a single purpose because they are connected with the means of knowing. The
irregularity rests only on a bogus reason.If the likeness must be of the right type, then the reason property, as determiner
of the likeness relation, must also be of the right type. The object under investi-
gation must be like objects which belong to a group, the typical members
of which have the target property. Va ̄tsya ̄yana implies that if the property in
question is a property shared by typical members of the class of cows, then the
reason property must be the class-essence cowhood.
In the last section, we saw that rational inference is linked with warranted
extrapolation, and we wondered what it was that made an extrapolation war-
ranted. The problem we have here is similar. We are now asking for the condi-
tions under which it is admissible to extrapolate a property from one object to
another. It appears to be admissible to extrapolate the property “rain-maker”
from one black cloud to another black cloud, but not from a black cloud to a white
cloud. It appears to be admissible to extrapolate the property “has a dewlap” from
one cow to. another cow, but not from one four-legged animal (a cow) to another
(a horse). There seems to be an order in the world of objects, a structure which
licences the extrapolation of properties in some directions, but not others. Objects
are grouped together on the basis of their likenesses and unlikenesses to one
another. The possibility of likeness-based and unlikeness-based rejoinders shows,
however, there are many different ways of making these groupings, many dif-
ferent metrics of likeness. So the problem is this – given some arbitrary property
we wish to extrapolate from one object to another, how do we decide which such
metric determines a standard for proper and warranted extrapolation? For an
extrapolation may be warranted under one likeness relation, but not another. So
not every inference of the standard pattern is permissible:
hinduism and the proper work of reason 435