Fregean Propositions, Unmereological Compositions, and Bradley’s Regress

Steven Nemes writes and I respond in blue:

I know you're in a bit of a mereology phase at the moment, but I figured I'd shoot this by you.

Mereology is the theory of parts and wholes.  Now propositions, whether Fregean or Russellian, are wholes of parts.  So mereology is not irrelevant to questions about the nature and existence of propositions.  The relevance, though, appears to be negative:  propositions are unmereological compositions, unmereological wholes.  That is to say, wholes that cannot be understood in terms of classical mereology.  They cannot be understood in these terms because of the problem of the unity of the proposition.  The problem is to specify what it is about a proposition that distinguishes it from a mere aggregate of its constituents and enables it to be either true or false.  No constituent of an atomic proposition is either true or false, and neither the mathematical set, nor the mereological sum, of the constituents of any such proposition is true or false; so what is it that makes a proposition a truth-bearer?  If you say that a special unifying constituent within propositions does the job,then you ignite Bradley's regress.  Whether or not it is vicious is a further question.  Richard Gaskin maintains the surprising view that Bradley's regress is "the metaphysical ground of the unity of  the proposition."  Far from being vicious, Bradley's regress is precisely that which "guarantees our ability to say anything at all."

For more on this topic, see my "Gaskin on the Unity of the Proposition," Dialectica vol. 64, no. 2 (June 2010), 265-277.  It is part of a five article symposium on the topic.

I am not sure if you believe in Fregean propositions or not. As for myself, I don't look favorably upon the idea of Fregean propositions because of the problem of Bradley's regress. (I am assuming propositions would be composite structured entities, built out of ontologically more basic parts, maybe the senses of the individual terms of the sentences that expresses it, so that the proposition expressed by "Minerva is irate" is a structured entity composed of the senses of "Minvera", "irate", etc.)

I provisionally accept, but ultimately reject, Fregean propositions.  What the devil does that mean?  It means that I think the arguments for them are quite powerful, but that if our system contains an absolute mind, then we can and must reduce Fregean propositions to contents or accuusatives of said mind.  Doing so allows us to solve the problem of the unity of the proposition.

By the way, what you say in parentheses is accurate and lucid. 

In your book, you offer a theistic strategy for solving the problem of Bradley's regress as applied to facts. I don't know that a theistic solution to the problem as applied to propositions works as smoothly because of the queer sort of things senses of individual terms of sentences are supposed to be. The building blocks of facts are universals, which are somewhat familiar entities; but the building blocks of propositions are senses like "Minerva" which are murky and mysterious things indeed. What the hell kind of a thing is a sense anyway?

A sense is a semantic intermediary, an abstract 'third-world' object neither in the mind nor in the realm of concreta, posited to explain certain linguistic phenomena.  One is the phenomenon of informative identity statements.  How are they possible?  'George Orwell is Eric Blair'  is an informative identity statement, unlike 'George Orwell is George Orwell.' How can the first be informative, how can it have what Frege calls cognitive value (Erkenntniswert), when it appears to be of the form a = b, a form all of the substitution-instances of which are false?  Long story short, Frege distinguishes between the sense and the referent of expressions. Accordingly, 'George Orwell' and 'Eric Blair' differ in sense but have the same referent.  The difference in sense explains the informativeness of the identity statement while the sameness of referent explains its truth.

Further, propositions are supposed to be necessarily existent; hence the individual building blocks of the propositions must also exist necessarily. But how could the senses expressed by "Minerva" or "Heidegger's wife", for instance, exist when those individuals do not? (This is the same sort of argument you give against haecceity properties conceived of as non-qualitative thisnesses.)

If proper names such as 'Heidegger' have irreducibly singular Fregean senses, then, as you well appreciate, my arguments against haecceity properties (nonqualitative thisnesses) kick in.  It is particularly difficult to understand how a proper name could express an irreducibly singular Fregean sense when the name in question lacks a referent.  For if irreducibly singular, then the sense is not constructible from general senses by an analog of propositional conjunction.  So one is forced to say that the sense of 'Minerva' is the property of being identical to Minerva.  But since there is no such individual, there is no such property.  Identity-with-Minerva collapses into Identity-with- . . . nothing!  Pace Plantinga, of course.

In the case of identity-with-Heidegger, surely this property, if it exists at all, exists iff Heidegger does.  Given that Heidegger is a contingent being, his haecceity is as well.  And that conflicts with the notion that propositions are necessary beings.  Well, I suppose one could try the idea the some propositions are contingent beings.

Are there any solutions to the former problem (which you've blogged and written about before!) you think are promising? Further, what do you think of the second problem?

Perhaps you think the second problem can be sidestepped by saying that "Heidegger's wife" is just shorthand for some longer description, e.g. "the woman who was married to the man who wrote a book that began with the sentence '…'". I don't know that it is so easy, because that sentence itself makes reference to things that are contingently existent (women, men, books, sentences, marriage…).

Yes,all those things are contingent.  But that by itself does not cause a problem.  The problem is with the notion that proper names are definite descriptions in disguise.  If the very sense of 'Ben Franklin' is supplied by 'the inventor of bifocals' (to use Kripke's example), then the true 'Ben Franklin might not have invented bifocals' boils down to the necessarily false 'The inventor of bifocals might not have invented bifocals.'  (But note the ambiguity of the preceding sentence; I mean the definite description to be taken attributively not referentially.)