[hist-analytic] RBJ's Proposal on analyticity

Bruce Aune aune1 at verizon.net
Sun Mar 8 13:58:52 EDT 2009


The positivists held that analyticity and necessity are co-extensional  
because they were convinced that necessities could be known only by  
means of conceptual analysis.  Rationalist and Kantian alternatives  
seemed unscientific and therefore unacceptable.  I don’t think any  
informed, reasonable philosopher con responsibly contend that the  
coextensiveness of analyticity and necessity is “obvious.”  In view of  
the plethora of arguments to the contrary, I think it can be defended  
only by a pain-staking argument, the sort I attempted in my recent book.

  I am aware of no fallacies in Kripke’s arguments for the truth of a  
posteriori necessities (as we may call them) or even for the truth of  
contingent a priori truths.  As for the former, Kripke never said, “an  
identity between rigid designators must be necessary.” What he did say  
was that a statement containing rigid designators would, if true, be  
necessary.  For him, necessity is not a relation between designators  
(at least generally) but a relation between things: it is a relation  
holding between each thing and itself.  But his views on the necessity  
of true identities did not depend on the notion of rigid designation,  
and it was inadvisable for me to have mentioned this notion in my  
email, since it always stimulates useless discussion.  The formula “(x) 
(y)(x = y à N(x = y)” is a theorem of quantified standard first-order  
modal logic, and its proof in no way depends on a doctrine of rigid  
quantification.*  (You can verify this by looking at any standard text  
on modal logic, e.g. the one by Hughes and Cresswell.) The relevance  
of the modal theorem to your thesis can be illustrated by standard  
examples.  One is this:  If water = H2O, then it is necessary that  
water = H2O.  The assertion “water = H2O” is not (most people will  
agree) true by virtue of meaning, and neither is “N(water = H2O).” The  
truth-value of the first can be decided only by empirical  
investigation, and the second can be inferred from the first by means  
of the modal theorem.  If the previous example troubles you, here is  
another: “The inventor of bifocals is Benjamin Franklin,” where “the  
inventor of bifocals” is used to pick out a certain man--the man who,  
as it happened, invented bifocals.”


* As I see it, Kripke introduced the notion of a rigid designator to  
disarm the sort of counter-examples people raised when he spoke  
informally of the necessity of true identities.  Those erroneous  
counter-examples invariably involve fallacies of equivocation, when a  
term denoting one thing in the antecedent of a conditional, denotes  
something else in the consequent.


   As for contingent (non-necessary) but a priori truths that can  
reasonably be considered analytic, consider the “The standard meter is  
one meter long.”  At one time (when there was a standard meter) this  
could be known a priori, by a priori inference from a meaning  
stipulation, but it is not necessary, since the standard meter, if  
heated sufficiently, would become appreciably longer than one meter.  
Please don’t bother to argue with me about these cases, which I am  
presenting quickly and informally.  If you continue to hold your  
present view, you better be ready to argue with others, though,  
because they are generally accepted as all sound.

  You say,” The only relevant part of the meaning of a sentence for  
determination of either analyticity or necessity is the truth  
conditions.  We know that a statement is analytic when those  
conditions tell us that the under all conditions the sentence is  
true.”  I can almost agree with this, but I think “truth conditions  
for an arbitrary sentence” is far more problematic than you suppose.   
What are the truth-conditions for “Bachelors are unmarried”?  If it is  
given by the biconditional, “’Bachelors are unmarried’ is true iff  
bachelors are unmarried,” what entitles us to say that the proposition  
expressed by “Bachelors are unmarried” is “logically necessary,” as  
you say? As I argue in my book, partial specifications of meaning (of  
the kind Carnap called “A postulates”) are, I think, vital for showing  
the analytic character of many sentences.  And it is by reference to  
such postulates and other semantical rules that we can infer the a  
priori necessity of many analytic truths.  Not all analytic truths  
express necessities.
  As I said before, one of the classic questions of epistemology is  
whether there are synthetic a priori truths.  To move from the vague  
idea of true by virtue of meaning to analytically true simply begs the  
question against a host of traditional arguments, which have to be  
considered critically and fairly.  You can’t ignore them.
Your counterexample to the no color-overlap principle overlooked a  
crucial feature of that principle.  This is that a thing or region of  
a thing can have no more than one determinate color.  If a blue object  
is azure, its determinate blueness is not different from its  
azureness: it is the same thing.  There is no problem with a thing  
having two generic colors, or a generic color and a determinate one  
belonging to that color-genus.
I stand by all the claims I made in my memo: I think your approach to  
analyticity is not promising as it stands.  My ideas about the subject  
are developed at length in my chapters and 3, and I can only suggest  
that you have a look at them.  That is really all I have to say at  
this point.
Best regards,

Bruce Aune
Email Hotline Coordinator

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://rbjones.com/pipermail/hist-analytic_rbjones.com/attachments/20090308/6ba20156/attachment-0002.html>

More information about the hist-analytic mailing list