FOM: Existential commitments in logic

[Ketland,JJ]

Dear Juliette

I think that Frege is simply pointing out to Husserl the well-known fact
(nowadays), which we teach to undergraduates, but which troubled generations
of thinkers from Aristotle to Frege, that

	"All Fs are Gs"		(x)(F(x) --> G(x))

does not imply the existential claim

	"There exists an F"	Ex(F(x))

Of course, we have

	(x)F(x) |- ExF(x)		(*)

in ordinary classical first-order logic, because we always consider
non-empty models with at least one element. In particular, Ex(x = x) counts
as a logical truth. Another reason is that

	F(t) |- ExF(x),		(**)

where t is any term. So, first-order logic says "there is at least one
object", and that allows logical theory (especially the rules for
instantiation and generalization on arbitrary terms) to be developed
*smoothly*. But this is all really a matter of convention, as Quine has
often pointed out. See:

Quine, WV 1993: "Comment on Orenstein", in R. Barrett and R. Gibson (eds.)
Perspectives on Quine (Oxford: Blackwell), pp. 271-272.

Quine says:

	"The exclusion of the empty universe from predicate logic is purely
a matter of 	efficiency. If you leave the empty universe out of account,
you gain a lot of 	convenient laws that you can use in your deductions
in applied logic and which 	would not hold if you included the empty
universe. .... If you want to allow for the 	possibility that your
universe of discourse might prove empty, you can still handle 	the empty
case separately. It is trivial, in that all formulas there admit of an
immediate decision procedure: just mark the existentials as false and the
universals 	as true and resolve out truth-functionally. So there is
nothing philosophical about the 	empty universe in predicate logic"
(p. 271). 

In contrast, natural languages contain terms (like "Pegasus", "God", "social
justice in the USA", etc.) which designate nothing. One could, in principle,
adjust first-order logic so that no existential statements count as logical
truths (and inferences like (*) and (**) are not valid). Any such resulting
system is called a "(existence) free logic". 
The early workers in this field were (I think) Lambert and van Fraassen.
See, e.g.:

Lambert, K. 1965: "Existential Import Revisted", Notre Dame J. of Formal
Logic 6: 135-141.
----	1967: "Free Logic and the concept of existence", Notre Dame J. of
Formal Logic 8: 133-142.
van Fraassen, B.C. 1969: "Presupposition, Supervaluations and Free Logic",
in K. Lambert (ed.) The Logical Way of Doing Things (New Haven: Yale
University Press).

Some of the ideas behind such free logics are discussed in:

Mark Sainsbury 1991: Logical Forms (Oxford: Blackwell), Chapter 4, Section
20
Stephen Read 1995 (I think): Thinking About Logic (Blackwell, I think).


Jeff Ketland

Philosophy, London School of Economics