[hist-analytic] A posteriori knowledge of necessary truths

Roger Bishop Jones rbj at rbjones.com
Tue Dec 21 04:26:32 EST 2010

On Saturday 18 Dec 2010 18:23, Baynesr at comcast.net wrote:

> The context I had in mind was the usual discussion of a
> posteriori knowledge of necessary truths.

That's an informal setting of context, but then to be able 
to reason soundly, you need something more precise to start 
In that informal context there will be many different points 
of view about vital things like the meanings of the key 

> Aside from
> that, I don't think I can say much, if anything, about
> context. I don't think logical constructions really
> supply a useful methodology. It's been tried in
> philosophy of language with results accruing only to
> insights not requiring the logical rigamorole. in the
> first place. In other words, system building of this
> sort is typically an "unnecessary" shuffle.

Well I'm not familiar with any attempt to formalise this 
part of the philosophy of language, which I think starts 
with Kripke's Naming and Necessity, which is a big step 
backwards methodologically in my opinion.

> This is not
> to say that I don't believe first order logic (with
> identity) is not useful, or even that a model theoretic
> semantics of modal notions doesn't supply some
> illumination into a certain limited area of interests,
> only that it is "brittle" and often capricious. So we
> differ methodologically. 

I didn't intend to suggest that first order logic would be 

You must of course understand my remark as contesting the 
methods which are dominant in these studies (and in most 
analytic philosophy) which I regard as unsound and 

If there are problems with whatever attempts have been made 
at formal treatment, then they should be a basis for moving 
on to improved treatments.
If you were in a position to offer a rationale why such 
methods could not benefit, together with some kind of defence 
of less formal methods, then that could be debated.
However, you probably don't want to do that!

> The first part I accept, since I don't see that it
> contradicts what I've said etc. and I think it is pretty
> much right: knowledge of meanings is a posteriori etc.
> That's fine with me.

Well a corrolory is that there are no a posteriori 
necessities.  Is that fine with you?

> But now the concluding part.
> Introducing an existential won't entail that the sentence
> in which it occurs is de re.

That is what I said, thinking I was disagreeing with you!

> For example '(John believe
> (Ex)Fx)' has a de dicto reading; a de re reading would
> allow for wide scope: '(Ex)(John believes Fx)'. Here the
> quantifier "reaches" from outside the belief context <g>
> into the belief context. Here we have a de re reading.
> Compare: 'There is someone John believes is a spy' and
> 'John believes there is someone who is a spy'. Quite
> different; both existentials, but the semantics differs
> (on the usual assumptions). I conclude with a point on
> rigid designation.

I'm afraid I couldn't follow these examples.

> You are right to say that the reference of a rigid
> designator is fixed by its meaning; but that it is a
> rigid designator is not fixed by its meaning. So even
> though we come by meaning by way of experience, we do
> not come by our understanding that a term is rigid by
> way of experience; that is a semantical feature of
> names, such as 'gold', one that has no correlative in
> experience. I cannot *know* a term designates the same
> thing in all worlds if all I know is that it designates
> in this one; I have to know something about the
> semantics of the term that many competent speakers know
> nothing about.

The fact that a name is a rigid designator is a fact about 
language, and it is a fact which affects the truth conditions 
of sentences involving that name.
In my book that makes it a fact about the semantics of the 
language (though moot whether there are any languages in 
with rigid designators).

All such facts in relation to natural languages are 
contingent, and known (if at all) a posteriori (except some 
conditional assertions).
If indeed we cannot know that a term designates the same 
thing in every possible world then there can be no rigid 

Roger Jones

More information about the hist-analytic mailing list