Professional Documents
Culture Documents
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted
digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about
JSTOR, please contact support@jstor.org.
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at
http://about.jstor.org/terms
Springer is collaborating with JSTOR to digitize, preserve and extend access to Philosophical Studies: An
International Journal for Philosophy in the Analytic Tradition
This content downloaded from 139.184.14.159 on Wed, 14 Sep 2016 19:41:22 UTC
All use subject to http://about.jstor.org/terms
ERMANNO BENCIVENGA
Philosophical slogans are usually very vague. Of course, they may have
precise meaning within the systems of the authors who first propose them
but their use often extends substantially beyond the boundaries of those
systems, and their meaning becomes correspondingly less precise. In spite
of their vagueness, however, or perhaps because of it, such slogans are usual
quite stimulating: most of the time basic intuitions are expressed through
them, and it is interesting to see to what extent the intuitions in question
are independent of any given system, that is, how much of them can be ca
tured in different systems.
One of the most important and controversial slogans of contemporary
philosophy is the Kantian dictum that existence is not a (real) predicat
Certainly, this dictum expresses a presystematic intuition (that one migh
not share, but whose importance for many a speaker can hardly be denied
that is, the intuition that saying that something or other exists is differen
from saying that something or other is green in a sense in which saying th
something or other is green is not different from saying that something o
other is circular. In the present context, however, I will not be concerned
at all with the relation between the above dictum and intuition, or for th
matter with any way of directly defending the dictum itself, nor will I be con
cerned with the interpretation of the dictum inside Kant's philosophy. What
will be concerned with is the meaning the dictum can have in contemporary
philosophy of logic, especially in connection with the formal systems and t
formal semanties that have been developed recently for (existential presupp
sition-) free logics. Such systems and semantics will justify my looking fo
alternatives to what I regard as the two main attempts at interpreting th
Kantian dictum within the above mentioned context, and will suggest quit
naturally one such alternative, which constitutes the positive contribution
of the present essay.
The first line of interpretation of the contrast between existence and oth
predicates I want to discuss stems from Frege's work,2 and can be described
roughly as follows. Existence is not a property of objects, as for example is
redness; it is a property of concepts, and as such cannot be correctly
predicate of (say) John or the present King of France. So the contrast in
question is a contrast of levels: existence is a second-order concept, while
redness and the like are first-order concepts.
Certainly, this interpretation is reasonable, so far as it goes; but - one
might say - it simply does not go very far. One might agree with Frege's
intuitions about conceptual levels and with his identification of a second-
order concept of existence, and still claim that this is not the whole story.
One might think that Frege gives a good analysis of existence as involved
in statements of the form
and still be convinced that he should also have allowed for statemen
(2) a exists,
(3) (a) Standard (that is, Fregean) logic contains an implicit assumption
of singular existence, that is, the assumption that all the singular
terms available in the language are denoting.
(b) But the presence of this assumption is a defect of standard logic,
because among other things it restricts the range of applicability
of the logic and blurs the distinction between the inferences to
which the assumption is really relevant and the ones to which it is
not.
This content downloaded from 139.184.14.159 on Wed, 14 Sep 2016 19:41:22 UTC
All use subject to http://about.jstor.org/terms
AGAIN ON EXISTENCE AS A PREDICATE 127
Of course, this argument is not going to impress the standard logician. For
one thing, he is by no means forced to admit the truth of premiss (b), as he
can deny d la Russell that there are any (genuine) singular terms which do not
denote, hence that there is a wider applicability to be gained or there are dis-
tinctions to be blurred, or claim d la Frege that the presence of non-denoting
singular terms in natural languages is a defect of such languages, to be elimina-
ted in the construction of a logically perfect language.5 So we might try to
reformulate the argument in the following way:
(5) The concept of being (equal to) a has instances (that is, it has
general existence).?
Thus singular existence has been after all correctly ruled out, as it is an
unnecessary notion. The assumption you are talking about is (or, more
cautiously, can be interpreted as being) nothing but the assumption that
certain concepts are instantiated, that is, an assumption of general
existence, and you can study it, and even explore the consequences of
negating it, without introducing singular existence at all."
This content downloaded from 139.184.14.159 on Wed, 14 Sep 2016 19:41:22 UTC
All use subject to http://about.jstor.org/terms
128 ERMANNO BENCIVENGA
(7) a is red
(8) The concept of being (equal to) a has red instances (that is, it
has general redness)
(9) Rx(x=a)
(10) R(ed)a.
(1 1) E!a
This content downloaded from 139.184.14.159 on Wed, 14 Sep 2016 19:41:22 UTC
All use subject to http://about.jstor.org/terms
AGAIN ON EXISTENCE AS A PREDICATE 129
is a necessary condition for (12) to have any truth-value at all. The same
holds if we replace '...is young' with other typical predicates, such as '...is
fat' or '...is tall', but when we come to '...exists', that is, when we take (13)
itself into account, the situation gets more complicated. For on the one hand
it seems that, when John does not exist, we can at least say that he does not
exist, that is, that (13) is false and
is true, which would mean that neither (13) nor (14) presuppose (13), and on
This content downloaded from 139.184.14.159 on Wed, 14 Sep 2016 19:41:22 UTC
All use subject to http://about.jstor.org/terms
130 ERMANNO BENCIVENGA
can come out true. In both cases, a contrast between existence and other
predicates is thus established, which one may describe by saying that
existence is not a (real) predicate. And furthermore, this contrast does not
depend on any paraphrase of statements like (13) in terms of general
existence. It allows one to take (13) exactly at its face-value, that is, as predi-
cating existence of John.
So far so good. Unfortunately, however, when we consider more closely
the (free) logics which have tried to embody Leonard's suggestions, we realize
that the above is more an account of a possible position on the subject than
of an actual one. For if taken to its ultimate consequences, the choice of such
a position would entail that practically no logical law A could be asserted
without adding to it the proviso
but no free logician has ever found such ultimate consequences attractive.
As a matter of fact, many existing free logics agree in accepting
(17) a=a
(without any proviso) as a theorem, and the agreement becomes almost
universal for tautologies like
or
This content downloaded from 139.184.14.159 on Wed, 14 Sep 2016 19:41:22 UTC
All use subject to http://about.jstor.org/terms
AGAIN ON EXISTENCE AS A PREDICATE 131
(21) a isP
and
would be true whether or not there is one and only one young fellow.14
Certainly, one could insist on saying that a statement containing non-
denoting singular terms is, if true, analytically true,15 but this claim would
not help us in the present situation. For what is relevant here is not the
analyticity of the statements but the analyticity of the predicates (supposing,
for the sake of argument, that we take such a notion seriously), and if "...is
winged' or '...is young' were to be judged analytic on the ground of their
occurring in the above statement then any predicate (or perhaps any non-con-
tradictory predicate) would. And thus the contrast we had established
between existence and other predicates vanishes. 16
Various morals might be drawn from this conclusion. The moral I want to
This content downloaded from 139.184.14.159 on Wed, 14 Sep 2016 19:41:22 UTC
All use subject to http://about.jstor.org/terms
132 ERMANNO BENCIVENGA
Let us see how this scheme applies to some of the examples mentioned
above. Self-identity is simple: the term 'a' occurring in (17) may be non-
denoting in a particular situation, and as a result (17) will lack a factual
truth-value in that situation, but in all altemative situations in which 'a'
is denoting (17) will be true, hence it will be (formally) true also in the situa-
tion above. Elsewhere I have described this procedure as a mental experiment,
and indeed a good way of putting the matter intuitively might begin as
follows: "Well, 'a' is non-denoting here, but suppose it were denoting...".
The same line of reasoning applies to (24), but with one main complica-
This content downloaded from 139.184.14.159 on Wed, 14 Sep 2016 19:41:22 UTC
All use subject to http://about.jstor.org/terms
AGAIN ON EXISTENCE AS A PREDICATE 133
(in case 'John' were also non-denoting) are not relevant for the evaluat
(24), and as a result (24) is, and (26) is not,"9 (formally) true.
The above treatment of (24) has an important moral: the structure
mental experiments can be made much more precise by the analysis o
But this analysis need not be (totally) explicit: a good deal of it may
presupposed in what we know (or we establish) about the possible de
tions of presently non-denoting singular terms. In other words, on
remember that the above is a general scheme, working in some gener
(e.g., with self-identity) but in more specific cases requiring a supply
mation to give us the results we want. Once this is understood, the
a satisfactory treatment of (22) and (23) is clear. If we supply the sc
with an adequate semantics of the terms 'Pegasus' and 'Zeus', our mental
experiments relative to them will be much more definite, hopefully so
definite as to exclude situations in which Pegasus is a cow or Zeus is a res-
taurant and to verify the statements in question, while probably leaving
indetenninate the truth-values of, say,
and
These few remarks should be sufficient to show that the above scheme
allows for any predicate to occur in true or false statements containing
non-denoting singular terms. This conclusion is encouraging, but to get
closer to the answer we are looking for it is important to make the following
observation. Even if in the scheme in question one has two senses of predica-
tion, a factual one and a formal one, the second sense is essentially parasitic
on the first one. It is by means of the factual truth-values it assumes in alter-
native situations that we decide the formal truth-value (if any) of a factually
truth-valueless statement: a formal truth-value is after all simply a factual
This content downloaded from 139.184.14.159 on Wed, 14 Sep 2016 19:41:22 UTC
All use subject to http://about.jstor.org/terms
134 ERMANNO BENCIVENGA
This content downloaded from 139.184.14.159 on Wed, 14 Sep 2016 19:41:22 UTC
All use subject to http://about.jstor.org/terms
AGAIN ON EXISTENCE AS A PREDICATE 135
This content downloaded from 139.184.14.159 on Wed, 14 Sep 2016 19:41:22 UTC
All use subject to http://about.jstor.org/terms
136 ERMANNO BENCIVENGA
(34) VxPxa
(35) Pa&E!a
University of Pittsburgh
NOTES
This content downloaded from 139.184.14.159 on Wed, 14 Sep 2016 19:41:22 UTC
All use subject to http://about.jstor.org/terms
AGAIN ON EXISTENCE AS A PREDICATE 137
presuppositions are of tremendous importance when one undertakes to apply the ab-
stract system to a concrete subject matter. Unless the field of application satisfies
the presuppositions, ...the application is invalid... . The situation is further complicated
...by the fact that not every presupposition is always relevant....The remedy is to make
the presuppositions explicit." (p. 50); "modern logic would be... improved if these
presuppositions could be acknowledged and symbolized, and the circumstances under
which they are relevant or irrelevant be discriminated one from the other." (p. 51).
Notice also that the symbolization that Leonard proposes for singular existence (and
that we will use here) is 'E!'.
5 The Russellian way out is well-known. Less popular is perhaps the Fregean one,
explicitly stated in 'On sense and reference' (in: Translations from the Philosophical
Writings of Gottlob Frege, by Peter Geach and Max Black (eds.), Oxford 1952, pp.
56-78): "Languages have the fault of containing expressions which fail to designate
an object (although their grammatical form seems to qualify for that purpose) because
the truth of some sentence is a prerequisite." (p. 69, italics mine); "A logically perfect
language should satisfy the conditions, that every expression grammatically well
constructed as a proper name out of signs already introduced shall in fact designate an
object, and that no new sign shall be introduced as a proper name without being secured
a reference." (p. 70, italics mine).
6 The qualification 'good' seems to be more appropriate in the case of something like
the Fregean way out mentioned in the previous note.
7 Of course, the standard logician will paraphrase (2) into (5) (if and) only if he
accepts a Frege-like analysis of general existence. Otherwise, he might propose such more
cautious paraphrases as 'There are things that are equal to a'. But this would not
change the substance of his argument.
8 In particular, Frege's account of the matter seems to be at least partly justified by his
acceptance of a thesis that many logicians today would not agree on: the thesis that all
singular existence statements are utterly uninformative. Cf. the following passage of his
'Dialog mit Punjer uber Existenz' (in: Nachgelassene Schriften, Hamburg 1969, pp. 60-
75): "In dem Satze 'A ist sich selbst gleich' erfahrt man ebensowenig etwas Neues uiber
das A, wie in dem Satze 'A existiert' ". (p. 70).
Mind 59 (1950), pp. 320-344.
Journal of Philosophy 69 (1968), pp. 136-52.
l I need hardly to point out that, whatever (real) predicates might turn out to be, they
should have little to do with the primitive predicates of any formal language. For the no-
tion that we are after should cover predicates of any degree of complexity, but formal
languages usually (with a few interesting exceptions) do not allow us to construct
compound predicates, but only compound formulas. Thus a more adequate formal
analogue of the notion we are interested in is the notion of an open formula, and on the
ground of this analogy we can reasonably say that (18) predicates something (namely,
Px v -Px) of a.
12 The most important bivalent semantics for free logics are van Fraassen's semant
classical valuations (for which see e.g. his 'The completeness of free logic', Zeitschrift
fur mathematische Logik und Grundlagen der Mathematik 12 (1966), pp. 219-34) and
Leblanc and Thomason's semantics of outer domains (for which see their 'Completeness
theorems for some presupposition-free logics', Fundamenta Mathematicae 62 (1968),
pp. 125-64).
13 This is the semantics grounded on the notion of an S-model, sketched in
quoted in the previous note.
14 See my 'Free semantics for indefinite descriptions', forthcoming in: Journal o
Philosophical Logic, and 'Free semantics for definite descriptions', unpublished.
'5 Such an insistence however would not be very reasonable, for all the abov
semantics but mine. For take for example van Fraassen's semantics of S-models.
This content downloaded from 139.184.14.159 on Wed, 14 Sep 2016 19:41:22 UTC
All use subject to http://about.jstor.org/terms
138 ERMANNO BENCIVENGA
Certainly, if Pegasus does not exist, there are S-models (for some set S) in which (22)
is true, but there are also S-models (for the same set S) in which (22) is false. And how
can this be consistent with the claim that when Pegasus does not exist (22) is, if true,
analytically true?
16 One might try to save this line of solution by first establishing that a statement
expresses a real predication if either it is analytic or it has existential presuppositions,
and then defining real predicates as I do at the end of this paper. I have nothing in
principle against this way out, and indeed one might conceive of my personal
contribution as the result of adding some qualifications to it. It is worth noticing,
however, that some qualifications are needed, in view of the ambiguity of the notion
of analyticity (which ambiguity, of course, is strictly connected with the problems
mentioned in the previous note).
17 See my 'Free semantics', forthcoming in: Boston Studies in the Philosophy of
Science, 'Truth, correspondence, and non-denoting singular terms', read at the 1978
Meeting of the Society for Exact Philosophy, and the papers quoted in note 14.
18 It might be useful to say explicitly that I am using the word 'statement' to designate
ambiguously statements of natural language and closed formulas of a formal language.
19 Of course, (26) is not true by now, but it might become true later, on the ground of
some analysis of the term 'John'.
20 It is easy to see that the procedure mentioned in the text admits of limit-cases in
which every statement containing some non-denoting singular term receives a truth-
value. Such limit-cases would give us of course bivalent semantics.
21 In my own development of such matters (e.g. in my 'Free semantics') I adopted
the first alternative.
22 In my 'Free semantics' and in the papers quoted in Note 14 I modified the above
scheme also with respect to identity (not, however, with respect to self-identity). This
modification, though not being strictly necessary, was in my opinion very natural (in
the sense that we can avoid it only by rather ad hoc hypotheses); hence I think I might
argue that a statement of the form 'a = b' (where 'a' and 'b' are different singular
terms) does not express a real predication. But this is a story for another time.
23 It may be worth noticing that a necessary condition for the reasonableness of this
definition is that, if any such substitution results in (a statement expressing) a real
predication, every such substitution should result in a real predication. But this
condition can be easily proved to hold for the scheme in question.
24 In my 'Truth, correspondence,...' and elsewhere I argued that in a very important sense
my scheme dispenses with non-existing objects. Thus when I use colloquially the expression
'non-existing objects' I prefer to put it between quotation-marks, to emphasize its
technical inappropriateness.
This content downloaded from 139.184.14.159 on Wed, 14 Sep 2016 19:41:22 UTC
All use subject to http://about.jstor.org/terms