COLIN MCGINN

REPLIES TO DISCUSSION ON LOGICAL PROPERTIES

Let me begin my response to my three commentators by thanking

them for their efforts: I found their comments generally helpful and
sympathetic, and commendably clear. If I disagree with many of
their points, that should not detract from my appreciation of them.
I won’t in what follows go into every nuance and detail of my
disagreements, instead sticking to points of larger significance; nor
will I do much by way of recapitulating their arguments.

Michael Glanzberg discusses my chapter on truth, which he

objects to on several counts. The first concerns my claim that
only truth has the disquotational property, despite some apparent
counterexamples. The example he brings up is the notion of
verification, particularly proof in mathematics: surely, if it has been
verified or proven that p, then p. It is of course perfectly clear that
verification is not the same as truth, on pain of denying unverifiable
truths – just as knowledge is not the same as truth. But the question
is whether the notion of truth is built into verification and proof as a
necessary condition, and hence is what powers disquotation. In fact,
I do advert to this issue, under the heading of justification, and I take
it that my reply to the alleged counterexample should be obvious,
namely that to verify something is to verify it as true. Something
counts as a proof only when it shows that the conclusion of the
proof is true. If we adopt a weaker notion of verification, as that a
proposition is shown probable by what verifies it, then we cannot
deduce the proposition, but only that we have non-conclusive
reason to assert it. So we need to assume, if we are to secure the full
strength of disquotation, that verification is of the conclusive kind,
i.e. there is no logical room for the falsity of the conclusion – but
then we have introduced the notion of truth. We can infer from “it
has been proved that p” that p only if we can also infer “it is true

Philosophical Studies 118: 453–461, 2004.

© 2004 Kluwer Academic Publishers. Printed in the Netherlands.
454 COLIN MCGINN

that p”. So I think that the concept of proof is not a counterexample

to my uniqueness claim.
Glanzberg is right to suggest that my main critical target is
the equivalence thesis, which regards “p” and “it’s true that p” as
synonymous. One of my arguments against this thesis is that in the
case of truth-value gaps the left hand side of the biconditional will
be false when the right hand side lacks truth-value. This is simply
because it is false to say of a proposition that is neither true nor
false that it is true. No doubt formal systems can be devised that
assign identical truth-values to both sides, but my point is that, on
the ordinary understanding of “true”, it is simply false to ascribe it
to what lacks the property of truth (consider ascribing it to a table).
My point here is not that the biconditional can never be read from
right to left; it is that it cannot be so read simply given the bare
content of the disquoted right. At the least we need to make the
entailment conditional on the existence of suitable truth bearers: if
we have it that p, and that there exists the proposition that p (or a
sentence expressing this proposition), then we can deduce that the
proposition that p is true. And of course we also need to take it that
we have a concept of truth available. This is all because the left
hand side is conceptually richer than the right hand side. It is not
that we can’t reasonably make these assumptions; it is just that they
are conceptual additions to the mere fact that (say) snow is white.
Hence the equivalence thesis is false. I don’t then think there is any
real disagreement between Glanzberg and me about the right to left
entailment.
When I speak of disquotation I am not to be taken literally, as
I note at the outset of my chapter, since propositions and beliefs
are not themselves linguistic (unless we make them so), though
they are representational in some way. When we get to the relation
between language and propositional content we enter the domain of
semantics, which I was avoiding. My concern was just with what
it means to say of a propositional content that it is true; I was not
concerned with assertion and the like. I see no problem in bringing
assertion into the picture, as Glanzberg does, but I also see no objec-
tion to keeping it out if one’s concern is simply with the meaning of
“true”.
 REPLY 455

Why is truth philosophically interesting? No doubt for many

reasons, but one of them, as I was concerned to stress, is that truth
is a property of a proposition that points beyond itself: when a
proposition has this property things have to be as they are stated to
be – and no other property has this consequence. I don’t see that
Glanzberg has given any convincing reason to contest this.

Gregory Fitch focuses on my chapter about existence. Contem-

porary philosophers tend to be so wary of non-existent objects that
it is hard to get them to see what such talk is really committed to
(and not committed to). Fitch slides readily into speaking of my
being “ontologically committed” to “there being certain entities”
such as unicorns. I myself would eschew all such loaded language,
which insinuates contradiction in the very notion of non-existent
objects. Let us speak instead, for the benefit of those steeped in
Quinean categories, of non-existent subjects – topics of discourse
that don’t exist, such as unicorns. So we can say that some of the
subjects of our discourse exist and some do not. And that is all
that should be said – forget talk of “ontological commitment” and
“entities”. Now I would say that, in the ordinary use of “refer”
(and “about”, “of” and associated locutions) there are subjects of
discourse to which we refer that do not exist – for how else can they
become subjects of discourse? It is sheer tendentious stipulation to
insist that “refer” be restricted to the existent subjects of discourse.
Similarly, in the book, I argue that the English word “some” does
not have existence built into it, so that we can say things like “Some
centaurs are virtuous”. Accordingly, “exists” is not a universal
property, contrary to Fitch’s suggestion, if that means that every
subject of discourse, i.e. object of reference, has it: and clearly we
do talk about unicorns, though they don’t exist. This, it seems to
me, is just dear old ordinary language, and the burden is on the
opponent to challenge it in some non-question begging way.1
Fitch wonders what I meant by “empty name” when I said that
“Vulcan” is one. I had hoped the context would make clear that I
meant “name with no existent reference” (which is what it is usually
taken to mean). I do of course hold that such a name has a non-
existent reference, since it is used to introduce subjects of discourse
that don’t exist.
456 COLIN MCGINN

I also think that some alleged names really do have no reference

of any kind, and Fitch’s case of “Charley” may be one such. Pure
nonsense words have no reference, not even to a non-existent object,
as when I say, out of the blue, “Zomwotoolies fly”. Clearly what
is needed here is some convention or tradition of understanding or
definition, but it would be difficult to make this precise (and there
will no doubt be borderline cases such as “slithy toves”).
I have no particular objection to distinguishing fictional names
from names born in error, such as “Vulcan”; no doubt names of non-
existent objects can come in many varieties. But I see no objection
here to anything I maintained, since clearly such names have one
important feature in common – they name non-existent objects (or
subjects); and this is the feature I was interested in.
Fitch wonders why I insist that non-existent objects like Vulcan
and unicorns necessarily don’t exist. Here I was following Kripke’s
famous discussion rather than proposing anything new; I refer
Fitch to this discussion.2 The basic point is just that many distinct
planetary bodies could satisfy the descriptions associated with
“Vulcan”, and there is no saying which of these would be the specific
planet ostensibly named by “Vulcan”. There is no specific possible
object of which it can be said that it might have been Vulcan.
Fitch turns next to modality. My point against the possible worlds
analysis of modal words is not that it fails to analyze modality in
a non-circular way; for the account I favor – the copula modifier
view – signally fails to do that. My point is purely semantic: not all
modal words can be given a translation into quantifiers over worlds
– notably “possible” in the phrase “possible world” (ironically
enough). Since I will be discussing this further in reply to Bricker,
I will be brief here. Does Lewis’s understanding of possible worlds
enable him to characterize them without using the word “possible”?
It may seem so if we focus on the idea that possible worlds are big
concrete particulars like the actual world: I and all my surroundings,
my various counterparts and all their surroundings, etc. But surely
these universe-sized concrete particulars have to be possible objects
– specifically, they have to consist of possible arrangements of
spatio-temporal parts. There is no such possible world, for example,
as one in which lions and zebras occupy exactly the same spatio-
temporal regions at the same time, since distinct objects cannot be
 REPLY 457

thus spatio-temporally coincident. Thus the concept of possibility

must be used to identify the range of quantification; which is simply
to say that the word “possible” in “possible world” is not redundant.
Nor should anyone expect it to be, given that reductionism about
modality is hardly on the cards. And, to repeat, I have no objection to
a semantic account of modal words that uses modal words – as mine
obviously does. My purely semantic point is that not all occurrences
of “possible” can be translated into the quantifier idiom, specially
those that occur in the phrase “possible world”. In other words,
we have not been told the logical form of every modal sentence.
By contrast, I claim, we can give the logical form of every modal
sentence by adopting the copula modifier theory – even though we
frankly use the words “necessarily” and “possibly” in stating that
theory (as a Tarskian theory frankly uses in the metalanguage the
very words it is interpreting in the object-language).
Fitch does raise a problem with my positive theory that has
puzzled me – how to account for uses of “possibly” in which the
property in question is not actually instantiated by the object, as
in “I might have been a plumber”. As I say in a footnote, we can
still suppose that the modal expression modifies the copula – as
in “I possibly-am a plumber” – but as it stands this introduces a
disagreeable asymmetry into the overall treatment, since we cannot
then say that all modal ascriptions are about modes in which objects
(actually) instantiate properties. However, it seems to me there is a
relatively easy way to restore the symmetry, which had not occurred
to me when I wrote the book: simply say that “I might have been a
plumber” means “My not being a plumber is contingent”. That is, if
an object that is not in fact F is possibly F, this is equivalent to its
being not-F only in the mode of contingency. And this sounds intui-
tively right, since what we are really saying is that, though the object
lacks the property, this lack is not a matter of necessity. Possible
F-ness is possessing not-F-ness in the mode of contingency. This
solves the problem and brings a satisfying unity to the account.
Finally, Fitch objects to my argument that the predicate modifier
view leaves open a logical entailment of modal sentences that the
copula modifier view does not. Here my point is not that modal
iterations are closed by the latter view and not by the former –
that would seem to me to be more like an objection to my view.
458 COLIN MCGINN

My aim was to evoke an intuition that favors the copula modifier

view, namely that if I say “Socrates must be a man” I am in effect
asserting that his being a man is something that he enjoys in
the manner of necessity – indeed, this just sounds like a windy
paraphrase of the original. But the predicate modifier account
cannot deliver this aspect of the meaning of the sentence, since it
limits the scope of “must” to the predicate itself, as distinct from
the predication: it says simply that Socrates has – non-modally –
the property of necessary manhood. To this one might reasonably
respond: “Well, you are telling me he has this peculiar property
of necessary manhood, but I want to know in what modal manner
he has the property of manhood and you haven’t told me that”.
No doubt no more should be made of this point than it deserves
– as an intuitive way to see how the two views differ and how
the copula modifier view hits the semantic nail squarely on the
head, while the predicate modifier seems to be hitting it somewhere
slightly off target. I rather suspect, in fact, that the intuitions that
have motivated the predicate modifier view are better served by the
copula modifier view – it is just that that view is less familiar as a
formal resource. In any case, my point was not about iterations of
modality, as I state in footnote 5.

Phillip Bricker takes himself to disagree with me quite sharply

over both existence and necessity, but upon closer examination I
don’t think he does. I fear that he begins his criticism by quoting
me out of context, when I am initially indicating what is meant
by a property. My intention here was to tie down the notion of
property sufficiently to form a real dispute between those who
think existence is a property and those who deny it; and clearly
both sides to this dispute are in need of some determinate notion
of that over which they are disputing. Since it is hard to arrive at
a non-circular definition of the notion of property, I resorted to
paradigm cases – the idea being that existence is more like these
paradigms than it is like, say, a quantifier concept or the concept of
an object or a proposition. Of course, existence is not claimed to be
in all respects like these paradigms; indeed, I am at pains to point
out some serious disanalogies with empirical properties. I therefore
have no principled disagreement with Bricker’s suggestion that
 REPLY 459

the existence property differs markedly from other properties

(just as moral properties, say, differ markedly from “descriptive”
properties). Just as identity is a genuine relation, analogous to other
relations, though also differing from them in important respects,
so existence is a property, analogous to other properties, though
differing from them in important respects. I in fact welcome the
disanalogies Bricker brings out; hence there is no real disagreement
between us. My only aim in the passage he quotes was to locate a
viable notion of property at the outset of discussion, not to insist
upon an analogy in every respect between the existence property
and (say) color properties.
Bricker raises an interesting skeptical possibility – that I might
be a fictional character who falsely believes he exists, as Holmes
believes falsely that he exists. As he says, this is clearly absurd,
but the question is why it is. It certainly appears to present a prima
facie problem for Descartes’ Cogito – for Holmes thinks and does
not exist! Nice little problem, I agree. However, I fail to see why
this is meant to be a particular problem for me, and why regarding
existence as a universal predicate would dissolve it. The skeptical
problem arises, if it does, for anyone, since it appears that we can
always coherently wonder whether we are fictional objects who
mistakenly take themselves to exist: Descartes hasn’t considered
and ruled out this kind of possible falsity of belief, in which fictional
characters are thoroughly immersed. And how would regarding
existence as universal property help? True, if there are material
objects they exist, but the question is whether there are any – that
is, whether our representations of them are of existent things. If
Bricker was right, Descartes could have skipped the Cogito and
simply asserted that he exists because everything does. Knowing
that everything exists is no help when the question is whether there
is anything!
I think that Bricker misinterprets my objection to possible worlds
semantics, along similar lines to Fitch: again, I am not objecting that
such a semantics is circular – since my own account is even more
clearly “circular”. My objection is purely to the semantic thesis that
every use of “possible” can be translated into a quantifier. Again,
too, I don’t agree that Lewis’s conception of worlds is non-modal:
the idea of a world as an aggregate of particulars must be taken to
460 COLIN MCGINN

be modal, since only some combinations will yield possible worlds

– it’s no good trying to combine lions at a given place with zebras
at that place (at the same time). The idea of an aggregate here is
the idea of how things can – possibly – be combined. Nor am I
arguing against the use of possible worlds in other areas, say in the
analysis of causation. Nor do I think there is any intrinsic objection
to characterizing them as possible. Nor even am I objecting to
stating truth conditions this way. My point, to repeat, is just that in
using the phrase “possible world” one has used a modal expression,
and this expression must be a quantifier according to the semantic
theory I am considering – but, I argue, it can’t be, and I don’t
see either Bricker or Fitch contesting my argument. (The relevant
question to ask here is whether the copula modifier theory can
account for this use, and how much this matters. Clue: “w is a
possible world” means “w possibly-is the actual world”).

My argument about excluding impossible worlds was not intended

to countenance them as real entities; it was to accentuate the point
that when we say “a world” we must mean possible world. That
is, “possibly p” must be taken to mean “in a possible world, p”
– or else we have not captured the force of the original: but then
the re-appearance of the modal word is blatant, and (I argue)
cannot be translated into a quantifier that does not have “possible”
attached to it. (As I remarked, the case is exactly analogous to
claiming that “exists” translates into the existential quantifier and
then being forced to admit that this quantifier must be read “there
is an existent thing such that”: that occurrence of “exists” cannot
be translated in the intended way.) So, even if the notion of an
impossible world were deemed incoherent (but is a world in which
water is H3O incoherent?), my objection would still apply, since
assertions of possibility logically imply that the thing in question
is not impossible. I was simply dramatizing this banal point by
saying that possible truth obviously cannot be truth in an impossible
world – it must mean truth in a possible world. In other words,
the addition of “possible” here is not redundant, despite the way
people nowadays say merely “true in all worlds”, and hence needs
to be captured by the semantic theory of modal expressions as
quantifiers over “worlds”. My invocation of impossible worlds was
 REPLY 461

to highlight the point that the notion of possibility had to be part

of the notion of world in order that we can stand half a chance of
explaining “possibly p” as “in some world, p”. This is not, I would
have thought, really a point over which much heavy weather needs
to be made.
Let me say in conclusion that these are certainly difficult issues
about which much more remains to be said, and I thank my
commentators for helping to focus some of the issues.

NOTES

1 I would recommend Terence Parsons’ Nonexistent Objects (Yale University

Press, 1980) as an antidote to the reflex “anti-Meinongianism” of contemporary
philosophers, which unfortunately I had not read when I wrote Logical Properties.
2 See Saul Kripke, Naming and Necessity (Harvard University Press, 1973),

Appendix.

Department of Philosophy
Rutgers, The State University of New Jersey
New Brunswick, USA
E-mail: [email protected]