V. Raspa, (ed.) Meinongian Issues in Contemporary Italian Philosophy, Meinong Studies, vol. 2.

, Ontos Verlag,
Frankfurt, 2006, pp. 119-159.

Mario Alai

SPEAKING OF NONEXISTENT OBJECTS

Summary

Frege captures the intuitive meaning and truth-value of some empty sentences, and Russell of some others; but many
such sentences are not satisfactorily analyzed by either one. A contemporary approach in the footsteps of these authors
suggests how to preserve the truth-value of all such sentences, but not their actual meaning. In my view Meinong’s
important lesson is that mention of nonexistent objects cannot be eliminated from a number of commonsensical
platitudes without changing their meaning. Yet, in spite of his partially misleading language (that prompted Russell’s
criticisms), I think he is not actually committed to an ontology nonexistent objects, but only of thoughts or concepts, as
Russell, or representations, as Husserl1.

Can we speak of nonexistent things? And can sentences about them be true? Or must they be false?
Or neither true nor false? The discussion on these questions involving analytic philosophers,
Meinong and Husserl, has been long and yielded many unsatisfactory answers. No answer seems to
be right across the board, but needless to say, the correct insights of each position agree with
whatever is right in each of the others.
We shall begin by considering the following examples of “empty” sentences:

(1) The present King of France is bald

(2) The golden mountain is golden
(3) Sherlock Holmes was clever
(4) Frank thinks of the golden mountain
(5) The round square does not exist
(6) The dagger is over my head (said by one who is having a hallucination)

1. Frege

According to Frege2, the truth-value of a sentence is a function of the referent of its terms (because
the referent of a complex expression is a function of the referent of its components, and the referent
of a sentence is its truth-value). Hence, since in each of the above sentences at least one term lacks
referent, none of them is either true or false: they lack truth-value. After all, in general, what we say
is true or false depending on the object we are speaking about; but if there are no objects to which
our words refer, there is nothing about which our assertion may be true or false. Empty sentences,
for Frege, are expressions endowed with sense but without a referent, just as the empty terms
occurring in them.

1
I thank Francesco Orilia, Venanzio Raspa, Vittorio de Palma, and especially Ernesto Napoli for illuminating criticisms
and suggestions on earlier versions of this paper, and Maria Grazia Severi for reviewing my English text.
2
Frege (1892).
Now, this seems to be correct in the case of (1), for upon hearing this sentence our natural
reaction would not be that it is false (we would say so if we knew that there is a king in France, now,
and that he has hair on his head), but rather that there is something wrong with the sentence, at least
as uttered today: it cannot be either true or false, precisely because there is nothing of which it might
be true or false. However, there are problems with this solution. Aside from the perhaps minor
embarrassment caused by giving up the principle of bivalence for a meaningful sentence, the utterer
of (1) might actually mean (although a less natural and less immediately understandable way)

(11) There exists a present king of France, and he is bald,

which is clearly false; or one could directly and explicitly utter (1 1): now, from his point of view
Frege could not explain why (11) is false (or for that matter, why sentences like “Polyphemous
exists” are false). Moreover, although concerning the truth-value of (2) and (3) we have different
pre-theoretical intuitions, some of them say that (3) is true, and (2) tautologically true; sentences like
(4) are regularly used in everyday language and understood as unproblematically true or false; (5) is
certainly true; and no doubt there is at least an interpretation in which (6) is false 3: but again, none of
these sentences could be either true or false, if Frege were right. So, there are at least some readings
of our expressions Frege seems unable to account for.
It might be objected that (2) and (3) cannot be true, since (atomic) sentences are true if and
only if the entities referred to by the subject terms instantiate the properties or relations referred to
by the predicate terms; therefore, no sentence can be true if one of its terms lack reference. But as
stressed by Meinong4, there is a natural way of speaking in common language in which we
truthfully attribute properties to nonexistent objects, and even to objects we know do not exist, as
when we say that Sherlock Holmes was clever, Achilles was fast, Santa Claus has a white beard, and
the like.
According to Ernesto Napoli5, when we assent to similar claims we are not actually accepting
them as true, but just pretending they are true: the convention behind writing fiction, telling legends,
etc., is that the author or the narrator pretends to be making true claims, and the reader or listener
pretends to be believing them. Convincing as this suggestion is, it cannot be the whole story; for,
when the author introduces the name for a fictional character (according to Napoli not actually a
name, but a pretended name), he must give some meaning to it. That is, the author must offer a
description of the character, elucidate the concept associated to the newly introduced name.
Therefore, at least part of what he says about the fictional character also counts as conceptual
characterization, as sort of implicit definition for the name; hence, it is analytically true (true by
definition), actually true. Thus, the claims that Sherlock Holmes was clever, Achilles was fast, Santa
Claus has a white beard, may be understood both de re, as concerning supposedly existent people,
and de dicto, as contributing to the meaning of the names, or as analyzing the concepts they express.
If understood de re they cannot be true nor false (although we can pretend they are true, for the sake
of the story, as Napoli claims); if understood de dicto they are true. The same happens for (2):
although it is not true when understood de re, it is true when understood de dicto.
3
These and similar intuitions are what Francesco Orilia calls “dati meinongiani di base” (basic Meinongian data): Orilia
(2002), ch. 9.
4
In Meinong (1902), and Meinong (1904).
5
Napoli (2000).
2. Russell

Unlike Frege, Bertrand Russell sees no possible truth-value gap: according to his doctrine in “On
Denoting”, the deep logical structure of any sentence including a definite description or a proper
noun (which according to him amounts to a shortened definite description), is either that there exists
exactly one thing, or that there exists nothing, fulfilling such a description. 6 Thus, the logical
structure of a sentence of the form “The such and such is φ” is: “There exists exactly one thing that
is such and such, and it is φ”. When applied to

(5) The round square does not exist,

this analysis accounts very clearly for its truth, for it becomes

(51) There exists nothing that is both square and round.

Thus, (5) does not assume any nonexistent thing, but says precisely there exists nothing having the
two properties of being square and round. The analysis of

(6) The dagger is over my head (said by one who is having a hallucination)

is also convincing, for it becomes

(61) There exists exactly one thing that is a dagger [and salient in the present context], and it is over
my head.

Hence, as shown by (61), (6) is in effect a conjunction of two false claims (for there is no dagger
salient in the present context, and in particular, no dagger is over my head); hence, (6) is false, just
as intuition says.
Actually, there are other possible readings of (6): for instance, the hallucination might cause
the speaker to believe that a particular real dagger known to him or to her (for instance, his or her
friend’s dagger), is presently pending on his or her own head. (6) should then be read as

(62) The dagger [i.e.: my friend’s dagger] is over my head.

Of course, (6) is false even in this reading, and Russell’s analysis respects this intuition, for it
interprets (62) as

(63) There exists exactly one thing that is a dagger [and belongs to my friend], and it is over my
head.

6
See Russell (1905a) and Russell (1919).
Since (63) is the conjunction of a true claim (the first conjunct) and a false one, it is false. However,
there are also legitimate readings (although possibly less common or less obvious ones) in which (6)
is not false. For instance,

(64) The dagger [the one I am seeing, whether real or not] is over my head [i.e., that is precisely the
position where I see that dagger as being].

In everyday language the verb ‘to see’ some times is understood objectively, as referring to a
relation between the subject and an existing object; at other times, it is understood subjectively, as
describing a mere subjective sensorial experience, independently of the existence of the object that is
apparently perceived. In (64) ‘to see’ has the subjective reading, hence one would say (6 4) is true, for
the subject does have the experience of seeing a dagger over his or her head. But it is not clear how
Russell would account for the truth of (64): certainly not by analyzing it as (61), which is false. At the
end of § 6 I will indicate two paraphrases Russell might give ((67) and (68)), none of which is correct.
Besides, Russell’s analysis fails in all the other cases, for it misinterprets sentences (1)-(4). To begin
with,

(1) The present king of France is bald

Becomes

(12) There exists exactly one thing that is presently king of France, and it is bald,

which is but a paraphrase of (11), hence false, while (1) is neither true nor false: as we saw, in fact,
(11) is a possible but secondary reading of (1), having a different meaning and a different truth-value.
As pointed out by Strawson in response to Russell, by bona fide uttering (1), we typically
presuppose the existence of (exactly) one present king of France, without actually asserting it.
Hence, we are not asserting anything false, but the failure of our presupposition makes us unable to
actually make an assertion, i.e. to say something either true or false 7. As already noticed, therefore, a
listener’s most natural reaction would not be that we said something false, but something bereft of
truth-value. Thus, Russell’s analysis does not make justice to the most natural reading of (1), but
only to a less natural and common reading, i.e. (1 1)8. For the same reason, Russell does not make
justice to a further possible interpretation of (6), namely

(65) The dagger [whose existence I presuppose, without asserting it] is over my head.

Moreover, there is an intuitive reading in which we must grant that

7
See Strawson (1950), § III. Also Frege (1892) maintained that when uttering a sentence we presuppose that its subject
term refers; but as we saw in § 1, this is not always the case: we are not presupposing this, for instance, when uttering
(2) and (3) de dicto.
8
In (1905a) Russell denies the intuitive evidence that (1) is neither true nor false, but the only reason he gives is “since
it is plainly false”. He seems to presuppose here his earlier demonstration that the falsity of (1) follows from his theory
of denotation, but the whole argument becomes then a petitio principii.
(2) The golden mountain is golden

is not only true, but even tautological (indeed, this is the only reading in which many of us would
utter (2)); but this is not accounted for by Russell, since he renders (2) as the false sentence9

(21) There exists exactly one thing that is both a mountain and golden, and it is golden.

Of course, like (1), even (2) might be understood with a different, implicitly existential meaning:

(22) [Implicit: there exists a golden mountain, and] the [aforementioned] golden mountain is
[obviously] golden,

which is correctly rendered by (21); but again, this interpretation is neither the only possible, nor the
most natural one. One might try to deny that there is any reading in which (2) is true, claiming that
what does not exist (as the golden mountain) cannot be golden (or of any other nature): in other
words, claiming that in order to have a property it is necessary to exist, and that attributing a
property is attributing existence. But as we have seen, we cannot deny that (2) is true de dicto.
Another objection might be that while (2) is literally false (or, according to a different account,
neither true nor false), the intuition that it is true, nay, tautological, is caused by understanding it as

(23) Every golden mountain is golden,

which in fact does not make any existential claim, and whose logical form is spelled out by the true
universally quantified sentence

(24) For any object x, if x is a mountain and golden, then x is golden.

In other words, one might claim that, as Husserl also suggests 10, a discourse apparently concerning
nonexistent objects is actually a sort of hypothetical discourse. But why are (23) and (24) true? Over
which domain of objects must their quantifiers range, if they are to be true? If we say that the
variable x ranges over existent objects, and that (2 3) and (24) are true of such objects, then we must
grant that also

(25) For any object x, if x is a mountain and golden, then x is not golden

is true (since there is nothing that is both a mountain and golden, hence the conditional is made
trivially true by the falsity of its antecedent). Since this is clearly not what we mean, we must then
admit that x ranges over possible objects, including nonexistent ones 11, and this is what both C.I.
9
In (1905a) Russell denies that ‘The round square is round’ (a sentence of the same form of (2)) is true, simply on the
basis that its falsity follows from his theory of descriptions. Once again (see footnote 8) he seems therefore to commit
petitio principii: instead of founding his theory on intuitive evidence concerning the examples, he asks the reader to
ignore intuitions in order to accommodate the examples to his own theory.
10
In Husserl (1894). I thank Vittorio De Palma for calling my attention to this paper.
11
In other words, although we may sometimes understand a conditional as a material conditional, which would make
both (24) and (25) true, if we accept (24) we are most likely understanding it as a proper implication, which makes (2 4)
Lewis and Meinong would say, for in their view (as we shall see below) our discourse domain
should include nonexistent (or “intentional”12) objects. But this assumption is rejected by Russell
and Husserl alike. According to Russell, logic should not admit unicorns any more than zoology
does, and we should not say that, for instance, Hamlet exists in Shakespeare’s imagination, for what
exists is only Shakespeare’s and his readers’ thoughts13. Equally, Husserl stresses that there is no
distinction between existing and non existing objects, but between representations occurring in true
judgements of the form ‘A exists’ and others occurring in true judgements of the form ‘B does not
exist’; there do not exist different worlds (the world of myth, the world of poetry, etc., and the real
world), but just one world, the real one, and many different representations 14. But why then are (23)
and (24) true, and (25) false? Actually, for the same reasons which make true also the following
sentences, which in turn might be cited as grounds by one who considers (2) to be true:

(26) If the golden mountain existed, it would be golden.

(27) It is necessary that a golden mountain is golden.

But why are these true? Again, they seem to refer just to merely possible or intentional objects, so
that Russell’s analysis cannot account for their truth.
In the case of

(3) Sherlock Holmes was clever,

it is even clearer that the objection according to which a nonexistent object cannot have any property
is misplaced: there is an obvious and very intuitive interpretation in which (3) is true: for instance, it
would be classified as true by the man in the street thoughtlessly answering a sudden question, or on
a quiz show with true/false answers, or filling out a reading comprehension exercise. This behaviour,
as we noticed, has two possible explanations: because people pretend it is true (although it is neither
true nor false); and because they know it is true de dicto. Yet, (3) is rendered as false in Russell’s
analysis:

(31) There existed exactly one thing that was an English detective, lived at 221b Baker Street, etc.,
and it was clever.

Again, (3) might also be understood as the implicitly existential

(32) Sherlock Holmes [implicit: existed, and he] was clever,

and in this reading it would be rendered correctly by (3 1); but what about the more obvious and
natural non existential reading, in which it is true? If we offered the counterfactual analysis

true and (25) false: but then, its semantics seems to require possible objects.
12
I will use here the term ‘intentional’ (also employed by Husserl in this context) to indicate all the nonexistent objects
we may speak or think about, regardless of their being merely possible (as the golden mountain) or even impossible (as
the round square); in other words, all the typically “Meinongian” objects.
13
Russell (1919).
14
Husserl (1894).
(33) If Sherlock Holmes had existed he would have been clever,

we would encounter the same problems raised by (26): counterfactuals, like modalities and universal
quantifications, seem to require reference to possible objects. One might also claim that (3) actually
means either of the following:

(34) Conan Doyle writes (or implies) that Sherlock Holmes was clever,
(35) Conan Doyle writes (or implies): “Sherlock Holmes is clever”.

But (34) could be naturally understood as claiming that there exists a man, Sherlock Holmes, of
whom Conan Doyle writes that he was clever, and in this sense it would be false; moreover, neither
(34) nor (35) are paraphrases of (3), which does not say anything about Conan Doyle. So, we must
keep (3) as it is; and since it is true, either (contra Russell) we may refer to nonexistent objects, or
(contra Frege), a sentence may have a truth-value even if its terms do not refer.
While (2) and (3) have possible implicitly existential readings (although not the most natural
and intuitive ones) which are correctly rendered by Russell’s analysis, nothing like that is available
for

(4) Frank thinks of the golden mountain.

One might try to analyze it in Russell’s style (4) as

(41) There exists exactly one thing which is both a mountain and golden, and Frank thinks of it.

But (41) is false even if (4) is true. It does not help to move the quantifier into the scope of the
propositional attitude verb15, as in

(42) Frank thinks that there exists exactly one thing which is both a mountain and golden,

for we often think of something without thinking that it exists, and even believing that it does not
exist. Nor can we analyze (4) as a case of direct speech, because

(43) Frank thinks “Golden mountain”

is meaningless: except in case of orders, exclamations, and the like, direct speech expresses a
propositional thought, with predication; hence it must consist in a (complete) sentence, not just a in
name or description as in (43). On the other hand, it is perfectly possible to say that a person has
thoughts concerning an object, as in (4), without specifying which propositional thoughts (if any)
that person has.
If we look at Russell’s remark that there does not exist Hamlet, but only Shakespeare’s and his
readers’ thoughts, it seems that the most natural strategy for him should be to read (4) as about
15
Or, as Russell would say, move ‘the golden mountain’ to a secondary occurrence: Russell (1905a).
Frank’s thoughts or the concepts they involve. In other words, he might quantify existentially on one
of those thoughts or concepts, and then try to characterize its relationship both to Frank, and to the
golden mountain. In the next section I shall review one such attempt, and argue that it is successful
only to a certain extent. Later on, we shall see that to the extent to which it succeeds, it is also
compatible with the interpretation of Meinong’s doctrine I suggest.
Summing up, Russell’s analysis seems to make justice to (5), to the most important reading of
(6), and to some secondary readings of (1), (2), and (3); but it fails for (4) (at least prima facie), for
the primary readings of (1), (2), and (3), and for a secondary reading of (6). In other words, Russell’s
account is correct only for sentences that make existential claims, explicit or implicit, affirmative or
negative; for all the other sentences, it either yields the wrong truth-value, or it cannot explain why
they have the truth-value they have (unless one admits the intentional referents Russell wanted to
avoid).

3. A neo-Russellian analysis

We have seen that a possible solution for sentences which are not straightforwardly rendered by
Russell’s analysis is to interpret them as counterfactuals like (2 6) and (33), or sentences on possible
objects like (24), or modalities like (27); all these interpretations, however, seem to imply Meinong’s
and Lewis’ nonexistent objects. But we have also seen that (2) and (3) are true when taken as
implicit definitions, or anyway as conceptual claims. In fact, it has been suggested that any
implication of nonexistent objects could be avoided by reading descriptions like ‘the golden
mountain’ and names like ‘Sherlock Holmes’, as actually referring to concepts, so that no
ontological commitment is needed except to concepts; this is similar to our earlier suggestion that (4)
might be considered as being about Frank’s thoughts, rather then the golden mountain. Thus, when
paraphrasing our examples as counterfactuals, universal quantifications, or modalities, one could let
the quantifiers range not over objects, but over concepts. A strategy of this kind is advocated for
instance by Francesco Orilia16, who sees definite descriptions as expressing denoting concepts,
themselves understood as second order properties. For instance, (3) is read as:

(36) Being clever is a property F such that there exists exactly one thing that is an English detective,
lives at 221b Baker Street, etc., and is F.

In symbols,

(37) [λF ∃1x (ED(x) & LB(x), …&F(x))](clever).

Thus, the description

(38) [λF ∃1x (ED(x) & LB(x), …&F(x))],

shortened in
16
See Orilia (2002), §§ 4.7, 13.5, 13.9, 13.11, 15.1; Orilia (2005).
(39)[THE[ED & LB & ….etc.]],

expresses a denoting concept, and ED, LB, etc., are the properties involved by such a concept, i.e., its
constituent components. Now, according to Orilia, the intuitive truth-value of empty sentences can
be accounted for by assuming them to be about such concepts, instead of meinongian objects. Hence,
(2) may be rendered as

(28) The concept [THE[GOLDEN & MOUNTAIN]] involves the property of being golden,

or, equivalently,

(29) Being golden is a property F such that there exists exactly one thing that is a mountain, golden,
and F.

(3) may be rendered as (36) or, equivalently, as

(310) The concept [THE[ENGLISH DETECTIVE & LIVING AT 221b BAKER STREET & …
etc.]] implies the property of being clever,

where a concept implies a property if the latter is either involved by it, or conceptually implied by
the conjunction of the properties it involves17. Moreover, (4) becomes, approximately,

(44) Frank thinks of the concept [THE[MOUNTAIN & GOLDEN]],

or

(45) There exists a concept involving the properties of being a mountain, and golden, and Franks
thinks of it.

This solution yields a clear answer to the question why (2 5), (26), (27) are true: they are true for
conceptual reasons. However, if one looks at how such denotative concepts are characterized in (3 6)-
(39) above, one notices that they imply the existential claim that there is something fulfilling the
description of Sherlock Holmes, or of the golden mountain, etc. Hence, these denotative concepts are
not actually an alternative to Meinong’s and Lewis’ intentional objects; on the contrary, they seem to
require intentional objects, thus running into the same difficulties that, as we shall see in the next
section, plague them.
Moreover, (28) and (29), (36) and (310), (44) and (45) are not faithful paraphrases of the
originals, for (2), (3) and (4) do not speak of properties or concepts, but of objects (although
nonexistent objects). For instance, if we look at (44) versus (4), a concept is not a mountain, and
thinking of the latter is not the same thing as thinking of the former. Suppose Frank is a three-year
old child: then he probably has the notions of a mountain and of gold, so that he can think of the
17
Orilia (2002), p.221.
golden mountain; but he probably lacks altogether the notion of a concept (especially of a concept as
complex as these denoting concepts). Thus, he is probably unable to think of the concept of the
golden mountain. In order to think of X one needs the concept of X only; but in order to think of the
concept of X, one needs both the concept of X and the concept of a concept, and a young child may
have the former but lack the latter. Thus, (44) may be false even if (4) is true, and one may
understand that (2) or (3) is true, without even understanding (2 8) or (36). In other words, as also
stressed by Husserl, when we think (or desire, or fear, etc.) an object X by means of a representation
of X, we are not thinking, desiring or fearing the representation, but X itself18.
The inadequacy of this approach can also be appreciated by considering the problems it
encounters in some other examples. For instance, it seems that (1) should be rendered as

(13) The concept [THE[PRESENT KING OF FRANCE]] implies the property of being bald,

or, equivalently,

(14) Being bald is a property F such that there exists exactly one thing that is both presently king of
France and F.

But (1) (today, in its more natural reading, and differently from (1 1)) is neither true nor false, while
(13) and (14) are false because they imply that the present king of France exists; moreover, even
leaving this aside, they are false because the concept of the present king of France does not imply
baldness. The same problem would affect sentences of the form of

(311) Sherlock Holmes had a cold on March, 3rd, 1891,

to be rendered in this fashion as the false sentence

(312) The concept [THE[ENGLISH DETECTIVE & LIVING AT 221b BAKER STREET & …]]
implies the property of having a cold on March, 3rd, 1891.

Thus, even the neo-Russellian approach does not yield the actual content of (1)-(4). However, I think
its strategy of focusing on concepts to account for discourse on nonexistent objects is basically
sound. This may be appreciated by observing that, if (2 8), (310), (44) and (45) did not refer to denoting
concepts, having an existential implication, but to plain concepts, without any existential
implication, they would be logically equivalent, respectively, to (2), (3) and (4): though having a
different meaning, they would be true exactly in the same possible worlds. In § 5 I shall further
elaborate on this sound strategy.

18
Husserl (1894); (1901); (1901/13), pp. 377-440.
4. Introducing nonexistent referents: Meinong’s ambiguity

It may be thought that a better solution is offered by Meinong’s remarks implying that there are
nonexistent objects, or objects endowed not with being, but with “quasi-being” 19, or that for any set
of properties (including reciprocally incompatible properties, such as round and square), there is
some object exemplifying those properties20; if so, it might be thought, terms may refer to
nonexistent objects (such as the present king of France, the golden mountain, Sherlock Holmes), and
even to impossible ones (such as the round square) 21. As a consequence, (1)-(6) would no longer be
empty sentences, and the truth of (2), (3), (4) and (6 4) would be easily accounted for by showing that
it is a function of the referent of their terms, just as Frege thought: in other words, that these
sentences do speak of something, and what they say is made true by the objects they are about.
On the other hand, the intuitive lack of truth-value of (311) would be explained by pointing out
that although it is not empty, it speaks of an incomplete object: since Sherlock Holmes does not
exist, it is not determined in all respects (in particular, as to his having or not a cold on March 3rd,
1891), hence sentences concerning respects in which it is not determined cannot be determinately
true or false. The lack of truth-value of (1) could also be explained in the same way, but even
Strawson’s explanation is compatible with Meinong’s approach: for according to Meinong there is a
present king of France, but he does not exist; and since (1) presupposes the claim that he exists, such
a presupposition is false, hence (1) lacks a truth-value. Again, the fact that the present king of
France, the golden mountain, Sherlock Holmes, the round square and the dagger do not exist, would
explain the falsity of affirmative existential claims like (1 1), (21), (31), etc., and the truth of negative
existential claims like (5).
A position similar to Meinong’s is taken by C.I. Lewis: every expression has, beside an
intension and an extension (i.e., the class of all things to which it correctly applies), also a
comprehension, the class of all coherently conceivable things to which it might correctly apply22.
This approach is supported by the following intuitions we have about expressions such as ‘the
present king of France’, ‘the golden mountain’, ‘Sherlock Holmes’, and ‘the round square’: it would
seem that
(i) they “stand for” something, they designate something, even if it does not exist;
(ii) each of them designate a different thing (hence, it is not the case that they designate the empty
class, or nothing at all);
(iii) I may call ‘Fido’ my dog, and twenty years later, when it does not exist any longer, I can
still use that name to designate it: the name is a designating one, it has a semantic counterpart, even
if its putative referent does not exist. It might be observed that it designates an object that once
existed (and the same for future objects); hence it refers, after all. However,
(iv)if I speak both of Fido, my real dog, and of Lassie, a fictional dog which I believe is real, I am
using those names quite in the same way, and it would seem they work in the same way: they play
the same “referential” role. The inexistence of Lassie does not seem to prevent me from using

19
See Meinong (1902), Meinong (1904) § 3.
20
This implication of Meinong’s claims has been formulated in analytic philosophy as “the principle of freedom of
assumption”: see Orilia (2002), pp. 86-88, and Orilia (2005).
21
Meinong (1904).
22
Lewis (1943-1944). See also Carnap (1947), § 16. His definition of comprehension excludes therefore contradictory
objects, such as the round square, which instead are admitted by Meinong.
‘Lassie’ in the same way as ‘Fido’. The situation is not much different if I use a name (say,
‘Pegasus’) to speak of an object which I know does not exist: no difference may be noticed in the
functioning of language or in my way of using it, whether I speak of Fido, or of Pegasus. Also, this
approach is supported by the intuitions we appealed to in ciriticizing Frege, at the end of § 123.
Thus, it would seem that any term has at least an intentional referent; and if in the real world
there exists an object identical to it, then this object constitutes a second referent for the term, the
real referent; otherwise, the term has just one referent, the intentional one. But this approach raises
serious problems:
(a) there is a classical distinction between the meaning and the referent of a term (Frege sees it as the
distinction between sense and referent; Carnap between intension and extension; etc.), and it is easily
understood as a distinction between a notional or conceptual level and a real or ontological level; but
obviously an intentional referent is notional, not real: so it becomes unclear what the difference
between meaning and reference is. Moreover, it may be easily seen that for any intentional object
there is a distinct notion, or concept, and vice versa; so, it might be argued that intentional referents
are a useless double for concepts24.
(b) According to Russell25, the assumption of intentional objects would imply contradictions, such as
the claim that the round square is both round (since it is round) and not round (since it is square).
Moreover, if for any set of properties (including reciprocally incompatible properties, such as round
and square) there is an object endowed with those properties, it follows that the existent round
square exists, which is impossible since contradictory objects cannot exist.
(c) It is obscure what such a third alternative between existence and mere inexistence (quasi-being,
or being without existing) might be, and the very assumption of such a third ontological status is
utterly implausible.
(d) According to Husserl26, this would result in the unnecessary and misleading doubling of the
object, a practice as pernicious as apparently plausible: there are not two objects, the intentional and
the real one, but only the latter.
The idea that there is always an intentional referent, and sometimes also a real one, may be
supported by the fact that the thought of an object is often accompanied, or even constituted, by its
mental image, and the same image is often associated with the object’s name or description: hence,
the image may easily be identified with the intentional referent. Moreover, this idea is analogous to
the ideistic conception of knowledge originated in the Middle Ages and prevailing in the XVII
century, according to which, since we know things through our ideas, we directly know only ideas,
while knowledge of the corresponding things is merely indirect.
But this ideistic gnoseology is wrong, because even if ideas (representations, concepts, etc.) are
the means by (or the way in) which we know, it does not follow that they are the object we know;
just in the same way, from the fact that we always see through our eyes, it does not follow that we
always see (primarily, or exclusively) our eyes 27. Similar objections are raised by Husserl with his
criticisms of the “doubling of objects”28: first of all, our expressions are not always connected to

23
See also footnote 3.
24
This observation is in the spirit, if not exactly in the letter, of Carnap (1947), § 16.
25
Russell (1905a), (1905b). See also Orilia (2002), p.110.
26
In Husserl (1894).
27
See Alai (1994), pp. 65-68.
28
In Husserl (1894), (1901/13), pp. 384-389, 436-440, (1913), pp. 206-207.
mental images (and we do not always think by means of mental images); secondly (as we already
noticed), even when present, these images are the means by which we refer to the object, not the
object we refer to; thirdly, if we feel the need of an intentional object to explain how a term refers to
an object, and if we think that the image may be such an intentional object, then the same problem
arises even for the image itself: how does it refer to its object?29 We cannot postulate a third
intentional object, or we would start a pointless infinite regress30.

5. A clearer treatment: speaking of nonexistent objects

If this were the end of the story, the truth of (2), (3), (4) and (6 4) would remain a mystery.
Fortunately, there is a better way to interpret Meinong’s remarks: better both because it is closer to
Meinong’s own intentions31, and because it solves our problems without raising the objections (a)-
(d) of § 4.
As a matter of fact, Meinong explicitly says that the claim that “there are objects that are not”
is unnecessarily paradoxical, and that speaking of “the being of the non-being”, or of a third status of
“quasi-being” is absurd and may ingenerate confusions. He then explains that he actually means that
objects are “beyond being and not being”, and that “being-so is independent of being”; but these
expressions, although they may suggest what he is actually aiming at, are not in themselves much
clearer than the earlier ones. He also says that the being of an “objective” does not need the being of
its object32, and this may also seem unclear, until one realizes that an “objective”, in his terminology,
is more or less a proposition33: he is then actually saying that the proposition that X is φ may be true
even if X does not exist. This shows that Meinong does not have in mind a duplication of objects, or
the existence of intentional objects; all he means is that we may speak of nonexistent objects, and we
may truthfully attribute them properties, or in other words, that there are true propositions about
them. It is in this sense that being-so is independent of being: for instance, we may truly say that the
golden mountain is golden, even if the golden mountain does not exist. Meinong’s expressive failure
consists in phrasing his doctrines as if they were ontological, while they are semantic, and in calling
them a “theory of objects”, while he is actually aiming at a theory of predication.
One may wonder in which sense true predication about nonexistent objects is possible, i.e.,
how a sentence containing an empty term may be true: for, if truth is a relation between language
and the world, a sentence cannot be true unless there is something in the world to which its subject
term(s) refer, and it has the properties or relations attributed by the predicate term(s). But before
answering this question, it should be pointed out that this insight – that the explanation of the truth-
value of our examples is to be sought within semantics, not within ontology – was clear to the
analytic tradition, in particular to Russell (in spite of his other misunderstandings), and to Husserl.
As we already noticed, according to Russell saying that unicorns exist in heraldry or in literature is a

29
Following Wittgenstein (1953), Putnam argues that the similitude between an image and an object does not, by itself,
make that image an image of that object: (1981), pp. 19, 56-58.
30
A problem Plato had already spotted with his “third men” paradox in the Parmenides, and Husserl raises again:
Husserl (1901/13), pp. 384-389, 436-440, (1913), pp. 206-207.
31
On this see also Chisholm (1982).
32
Meinong (1904), § 4.
33
For instance, see Raspa (2005) and Poli (2005).
ridiculous loop-hole, and what exist are only pictures or verbal descriptions 34; and according to
Husserl, one should not speak of intentional objects, but of representations, some of which can be
part of true existential judgements, and some can be part of false existential judgement (or of true
negative existential judgements)35. In the terminology of analytic philosophy, that is to say that there
are concepts (or terms) figuring as subject of true existential propositions (or sentences), and
concepts (or terms) figuring as subject of false existential propositions (or sentences). In other
words, there are concepts (or names, or descriptions) of existent or of nonexistent objects, and the
concept (name, description) of a nonexistent object may figure as subject of true non existential
propositions (sentences).
How then can we have true predication about nonexistent objects? How can sentences be true
when their subject terms do not refer? First, negative existential claims such as (5) can be true only if
their subject term does not refer. In such cases one may analyze away the empty term, as Russell
does, showing that it does not actually occur in the deep logical form of the sentence, or one can
hold that the sentence is not actually about an object, but about a concept, of which it says that it is
not instantiated; but then it must be granted at least that the superficial subject term in the sentence
does not refer. Secondly, truth is not always a words-world relation: some truths are conceptual or
definitional truth, as (2) and (3) in at least in some possible reading. In such cases, even if the subject
term does not refer, the sentence is true if the property or relation attributed by the predicate term is
part of (or implied by) the meaning of the subject term. One may also point out that the sentence is
actually equivalent to a metalinguistic sentence, whose subjects exist, for they are the term itself and
a concept, and the sentence says that the latter is (part of) the meaning of the former; but even if so,
the subject of the original sentence does not exist. Thirdly, a sentence like (4) is not actually about
the golden mountain, but about a mental act of Frank’s, involving the concept of the golden
mountain: it says that Frank bears a certain relation (of grasping, or of dwelling upon) to that
concept. Yet, grammatically, the term involved is ‘the golden mountain’ (not ‘the concept of the
golden mountain’), and it lacks reference.
Thus, it is true that we speak of nonexistent objects, in the sense that at least the grammatical
subject term may be involved in a true predication even when lacking reference. But is there any
philosophical interest in such a claim on the grammatical subjects of our sentences? I think so,
because language works in such a way that some claims on concepts are most naturally expressed by
speaking in terms of objects, just as it happens with the de re reading of (2) and (3), with (4) and (5).
Also in (64) a claim on perceptions is expressed as a claim on an object. This is understandable, since
psychology and philosophy of language explain that both philogenetically and ontogenetically
concrete objects are the primary referents of our speech, because of their intersubjective nature 36;
only at a later time, by a process of abstraction, we come to focus on concepts (or terms, perceptions,
etc.), and in the beginning we may convey our claims on such matters only through the vocabulary
of objects. Moreover, as we shall also see in the following, in order to make it explicit that one is not
speaking of objects but of concepts (or terms, perceptions, etc.), we must use a metalanguage, or
anyway introduce new terms and concepts (e.g., the concept of concept, of term, of visual image,
etc.). Hence, Meinong’s point that we speak of nonexistent objects, on the one hand does not mean

34
Russell (1919).
35
Husserl (1894).
36
For instance, see Quine (1960), I § 2.
that we actually refer to such objects, or that they have any form of existence, because true sentences
in which a nonexistent object is the grammatical subject (like (2) and (3), (4) and (6 4) may be turned
into logically equivalent sentences in which the nonexistent object is explained away (like (2 12),
(313), (46) and (69); on the other hand, Meinong calls our attention on the fact that sentences of the
latter type do not have the same meaning as the corresponding sentences of the former type, and
cannot supplant them in our speech.
Thus, the truth-value of a sentence is not, in general, a function of the referents (understood as
those existing objects that are the semantic counterpart of terms) of its terms: if it were, for instance,
(2), (3), (4) and (5) should be neither true nor false. In other words, some sentences do not
presuppose the existence of referents for their terms; so, even if their subject terms lack reference,
they can be true (when making a negative existential claim or no existential claim at all) or false
(when making an affirmative existential claim). By the way, if actually truth-value were a function
of referents, co-referential expressions should be interchangeable salva veritate in all contexts, but
we know they are not (for instance, in indirect and modal contexts).
Actually, although Frege held that the truth-value of a sentence is function of the referents, he
was able to cope with the above problems by understanding reference (bedeutung) as something
that, in a very peculiar way, changes with the context: in indirect and modal contexts expressions do
not have their ordinary referent (an object for names, a function from objects to truth-values for
predicates), but an indirect reference, namely, their ordinary sense37. This works, of course, because
sense is stronger than reference (sense determines reference but not the other way out), and identity
of sense warrants substitutivity salva veritate in every context. And this works also for (2), (3), (4),
(5), and (64), if they are interpreted as concerning concepts, for Fregean senses are concepts, or
something very similar38. But then the most straightforward solution should be acknowledging that,
in general, truth-value is a function of the concepts expressed by terms (i.e., of Frege’s senses) not
of reference. In fact, this is the solution I am going to suggest, although, in order to avoid the
particularities of Frege’s characterization of sense and related problems, I shall rather speak of the
truth-value of a sentence as a function the meanings of its terms (as well as, obviously, of the
logical structure of the sentence and of the way the world is), and further identify meanings with the
concepts expressed by each term. This, in turn, will explain why sentences including terms which
have meaning but no referent may still be true or false.
When speaking of meaning, here, I have in mind something similar to Frege’s sense, in that it
is the cognitive content of our expressions, what we understand when we understand an expression,
and the way in which it identifies its referent. In addition, however, I identify the meanings of non-
logical terms with concepts, and the meaning of sentences with propositions. A concept, in turn, is
the abstract idea of an individual, or a kind, or a property or a relation 39, and a propositions is the
abstract idea of a state of things). There may be different concepts of the same thing (e.g., the
concept of the morning star, and the concept of the evening star), and so terms with different
meanings may have the same referent. A concept is the cognitive content of a psychological idea or
mental state, it may be common to different subjects and different mental states, it is non-spatial and
non-temporal. A concept or a proposition, therefore, may be the cognitive content common to an
37
Frege (1892); see also Carnap (1947), § 28.
38
Although they are not what Frege himself calls concepts, i.e. functions from objects to truth-values: see Frege (1891).
39
This is another difference from Frege’s, for whom concepts are functions from individual to truth-values: see Frege
(1891).
expression (the expression of which it is the meaning) and to a mental state. Concepts are structured,
so that a concept in general has other concepts as constituents (for instance, the concept of star is a
constituent of the concept of the morning star) and concepts imply other concepts (for instance, the
concept of star involves that of heavenly body). (In what follows, the claim that a concept C involves
(or implies) a property P is to be read as meaning that the concept of P is a constituent of (or implies)
C). Finally, it is not necessary here to address the various problems concerning the nature of
concepts, such as whether a concept is always sharply defined, or may be vague or fuzzy, etc.
Although for particular purposes concepts may be represented as sets or functions, we do not need to
assume that they are sets or functions.
I assume that the meaning of some proper names (at the very least those of nonexistent entities)
is the concept expressed by the (possibly non sharply delimited) definite description associated to the
name; for instance, the meaning of ‘Sherlock Holmes’ is the (possibly vague) concept of the English
detective who lived at 221b Baker Street, was friend of dr. Watson, etc40. If, as Kripke claims41, some
or most proper names have no analytic association with any definite description, but are linked to
their referent by a direct socio-causal chain, then I assume that the meaning of such a name N is the
concept expressed by a definite description like “The individual to which N is linked by its socio-
causal chain”42.
Thus, let us stipulate that a term designates an object (whether existent or nonexistent) when its
meaning is the concept of that object, so that for any name or definite description ‘ _ _ _ ’ we may
correctly say: « ‘_ _ _ ’ designates _ _ _ », and also : « ‘_ _ _ ’ designates + + + », if the meanings
of ‘+ + +’ and ‘ _ _ _ ’ are concepts of the same object. Thus, as long as a term designates something
(i.e., as long as it has a meaning), it has at least the ability to refer: if the designated object exists, it
refers to it, and if it does not exist, the term still designates it (i.e. it still has the ability to refer to that
object). When in everyday speech we happen to say, for instance, that the Italian description ‘La
montagna d’oro’ refers to the golden mountain, or that ‘Polyphemous’ refers to a mythological
character, etc., in such cases ‘refers’ has obviously the same meaning I just stipulated for
‘designates’. (As a consequence, when saying that the truth-value of sentence is not a function of the
reference of its terms, but of their meaning, we may also say that it is a function of their
designation). Husserl says something similar: that a representation has a real object means that it
represents an object and the object exists; that a representation has an intentional object simply
means that it represents an object (without any further implication on its existence)43.
If one wishes, one may still speak of terms as having intentional referents, but neither the
concept of reference nor the existence of intentional objects must be taken literally. ‘Referring’ and
‘designating’ differ in the same way as ‘finding’ and ‘seeking’: the former implies the existence of
its object, the latter does not. Hence, this account avoids the objections (a)-(d) raised in § 4 against
the “ontological” interpretation of Meinong’s objects: (a) ‘designating’ does not introduce any third
semantic level beyond meaning and reference, for it is entirely defined in terms of meanings (i.e.,
concepts); we need not abandon the meaning/reference dichotomy, understood as a
40
See Searle (1958) for a detailed treatment of the vague character of proper names meanings. In this particular
example, ‘dr. Watson’ and ‘221b Baker Street’ are themselves names whose meaning is a concept expressed by an
associated description.
41
Kripke (1972).
42
This may be applied also to proper names of natural kinds, if, as Putnam claims, the non-descriptive theory of
reference applies to them, too. See Putnam (1973), (1975).
43
Husserl (1894).
notional/ontological dichotomy; and there is no duplication of the role of concepts, for designating
just is having a meaning, i.e., expressing a concept. (b) Since nothing we say in this way implies that
the designated object exists, it would not a problem if contradictory statements were true of it, since
the principle of non contradiction holds only for what is real. But besides, we shall see in the next
section that in this way Russell’s contradictions cannot be derived anymore. (c) No obscure third
status between existence and non-existence is literally introduced, nor (d) any intentional object is
literally introduced beside the real one.
Carnap is also close to this solution, when he writes that Lewis’ and Meinong’s distinctions
(e.g., between real and non real but possible objects, or between possible and impossible objects) are
important, but they can be accounted for by speaking of terms or concepts: for instance, by saying
that the term (or concept) ‘unicorn’ is empty, or that the term (or concept) ‘round square’ is
necessarily empty44.
Summing up, we have seen that whenever empty sentences make or imply an existential claim
on nonexistent objects, as (11), (22), (32), (41), (5), (61), (63), they are either true (if the claim is
negative) or false (if the claim is affirmative), and this is correctly explained by Russell. When they
neither make nor imply, but presuppose an existential claim, they lack truth-value, and this is
correctly explained by Frege and Strawson. Finally, empty sentences that neither make nor imply nor
presuppose existential claims, such as (2), (3), (4) in their most obvious readings, and (6 4), may be
true or false because truth-value is not a function of referents but of meanings (or, equivalently, of
the designated objects); they may also lack truth-value, when meaning is not determined as to the
relevant property, as in the case of (311).
This proposal agrees with the correct insight of the neo-Russellian approach, i.e. the idea that
the intuitive truth-values of empty sentences may be accounted for by understanding them as
somehow concerning concepts, and that they are logically equivalent to sentences referring to
concepts. But as we noticed in § 3, they cannot be paraphrased simply by substituting terms referring
to concepts to the otherwise empty terms, for this would yield categorial mistakes like

(44) Frank thinks of the concept [THE[MOUNTAIN & GOLDEN]]

The same strategy, if consistently followed, would also yield

(15) The concept [THE[PRESENT KING OF FRANCE]] is bald,

(210) The concept [THE[GOLDEN & MOUNTAIN]] is golden,

etc., which are absurd because concepts cannot be bald, golden, etc. Although the sentences we are
discussing somehow convey conceptual information, their terms do not actually refer to concepts,
they are empty terms. It is only through a complete reformulation of such sentences that we can get
logically equivalent (not synonymous) sentences whose terms refer to concepts.

44
Carnap (1947), § 16. See above, § 3.
For instance, a person having a propositional attitude concerning X is not speaking or thinking of
the concept of X45, but just of X. However, he or she is able to do this by means of the concept of X.
More precisely,

(α) speaking of X

is

(β) using a word designating X, i.e. a word whose meaning is the concept of X;

analogously,

(γ) thinking of X

is

(δ) dwelling upon the concept of X (activating it in one’s mind, moving it to one’s working memory,
or, as Fodor might put it, placing it into a particular mental attitude “box”46).

In all such cases, if X exists one is referring to it, and if it does not, one is still designating it. We
may speak of nonexistent objects in the sense that we can designate them, or that our terms keep
their referential ability even when they actually have no referent. This, again, seems to be a faithful
account of the usage of terms like ‘speaking of’ and ‘thinking of’ with respect to nonexistent objects
in common language. Thus, we may best interpret Meinong’s doctrine that “being-so is independent
from being” as the claim that (precisely for the just given reasons) one can speak of X and say that X
is this and that, without either asserting or implying that X exists.
To be sure, (α) and (γ) say the same thing of (i.e., they are logically equivalent to),
respectively, (β) and (δ); however, they do not say it in the same way (i.e., they have different
meanings), and a speaker might not be able, in general, to appreciate their equivalence. (β) and (δ)
say in a sort of theoretical metalanguage what (α) and (γ) say in the object language. For instance,
who utters (4) might be unaware that he or she is saying something logically equivalent to

(46) Frank dwells upon the concept of the golden mountain.

Hence, it is important to stress that commonsense expressions of the form (α) and (γ) are correct and
we should go on using them, even in the awareness that by using them we are committing ourselves
to nothing more than what is required by (β) and (δ). Meinong’s chief merit was to uphold this literal
correctness of the commonsensical way of speaking, even if he was not altogether clear on the
ontological assumptions it carries along. The analytic tradition, on the other hand, was right in
maintaining that correct speech cannot in any way commit us to an ontology of nonexistent objects,

45
This would be the case if, as Frege holds, the ordinary sense of a term became its referent when the term occurs within
a propositional attitude.
46
See Fodor (1975).
but at most of concepts, meanings, and the like; however, it has not been able, in implementing this
insight, to save the exact meaning of expressions of the forms (α) and (γ).

6. Making the ontological commitments of (1)-(6) explicit

The above considerations can be confirmed by a final review of how the sentences (1)-(6) might be
rephrased in order to make their ontological commitments explict.

(1) The present king of France is bald,

intuitively lacks a truth-value (when uttered between 1792 and 1815, or after 1848). We saw that
even if we assumed nonexistent objects, it would still be neither true nor false, since the present king
of France is an “incomplete” object. But the thesis that there are nonexistent objects is ontologically
misleading: if ‘are’ plainly meant ‘exist’, it would imply the truth of the false sentence

(11) There exists a present king of France, and he is bald,

and we know this is not what Meinong means. But if ‘are’ is to mean something different from
‘exist’, this thesis implies that there is an unacceptable third ontological status between existence and
non-existence. Thus, unless by (1) the speaker actually means (1 1), the correct explicit rendering of
(1) is

(16) The one thing [whose existence is here presupposed, but not asserted] that is presently king of
France is bald,

which lacks a truth-value because, as Strawson says, a sentence presupposing a false existential
claim cannot be either true or false. On the other hand, if one meant (1) as the false de dicto reading

(17) The present king of France as such is bald,

his or her actual ontological commitments would be spelled out by the equally false

(18) The concept of the present king of France involves the property of being bald.

Thus, Meinong’s teaching is not much relevant to (1), but rather to the following examples, in which
existence is neither asserted nor implied nor presupposed:

(2) The golden mountain is golden

is most naturally read as the de dicto

(211) The golden mountain as such is golden,

and in this reading it is logically equivalent to (23), (24), (26) and (27); moreover, the doubt that any of
these formulations commit one to possible objects may be dispelled by noticing that they all are
again logically equivalent to

(212) The concept of the golden mountain involves the property of being golden

(quite similar to Orilia’s (28), but not identical, since it carries no existential implication). But since
(212) does not have the same meaning as (2) (which does not mention concepts), the original meaning
is correctly explained by saying that (2) speaks of the golden mountain; and since the golden
mountain does not exist, it is also correct to say that we speak about nonexistent objects. The same
happens with

(3) Sherlock Holmes was clever:

we would normally utter it in the de dicto way, as a sort of conceptual analysis which neither implies
nor presupposes the existence of Sherlock Holmes. Thus, the ontology underlying (3) would
become explicit in

(313) The concept of the English detective living at 221b Baker Street, etc., implies the property of
being clever

(analogous but not identical to the neo-russellian (310); but the meaning of (3) is more faithfully
given by saying that it describes a fictional (nonexistent) character. On the other hand, if one said

(311) Sherlock Holmes had a cold on March, 3rd, 1891,

the case would be similar, but not identical to (1): here, in fact, a sincere utterance might be
grounded on either of two different beliefs: that Sherlock Holmes existed (just as a sincere utterance
of (1) would be grounded on the belief that there exists a present king of France); or that Conan
Doyle (or whoever wrote the stories about Sherlock Holmes) stated or implied that he had a cold on
that day, so that the fictional character Sherlock Holmes has this property. In the first case, the
speaker might understand (311) either as implying the existential claim (and then (311) would be false
and spelled out by a Russellian paraphrase similar to (11)), or as presupposing it (and then (311)
would be rendered as

(314) The one thing [whose existence here is not asserted, but presupposed] that is an English
detective, living at 221b Baker Street, etc., had a cold on March, 3rd, 1891,

which is neither true nor false, just as (16)). If, on the other hand, the speaker utters (311) simply
because he or she believes that Conan Doyle or whoever wrote so, he or she means something false,
equivalent to
(315) Conan Doyle writes (o implies) that Sherlock Holmes had a cold on March, 3rd, 1891,

or

(316) The concept of the one English detective living at 221b Baker Street, etc., implies the property
of having a cold on March, 3rd, 1891.

But again, both (315) and (316) would correctly represent the ontological commitments of (3 11) as
understood by the speaker, but not its meaning: (315) is a metalinguistic claim, and (316) mentions the
concept of Sherlock Holmes, while (311) is an object language claim, and mentions Sherlock Holmes
himself. Thus, (311) has a conceptual or metalinguistic content, but unlike (3 15) and (316), it expresses
it in a vocabulary of objects. In this sense, it speaks of a fictional (Meinongian) character.

(4) Frank thinks of the golden mountain.

In accordance with Meinong and commonsense we explain (4) as saying that Frank thinks of a
fantastic (nonexistent) object; but this does not commit us to the existence of such an object, for the
content of (4) is spelled out without any reference to it by

(46) Frank dwells upon the concept of the golden mountain

(similar but not identical to Orilia’s (44)).

(5) The round square does not exist.

As it is an explicit existential claim, it is correctly rendered by Russell’s analysis (51). Russell

thought that sentences like (5) contradict Meinong’s assumption about nonexistent objects, and it
would be so if the assumption were that such objects possess some ontological status. But as we saw,
Meinong’s point is just that those objects are something we can speak about. In (5) we designate (as
opposed to refer to) the round square, and in so doing we say that it does not exist; moreover, what
we say is true, because a negative sentence as this one is true when the object designated by the
subject (the round square, in this case) does not have the property designated by the predicate
(existence, in this case). The reading I am suggesting shows that it is all right if, speaking of the
round square, we can derive both

(52) The existent round square exists

(a tautology, just like ‘the golden mountain is golden’), and

(53) The existent round square does not exist

(which is certainly true, because nothing having incompatible properties can exist). In fact, (5 2) and
(53) do not contradict each other, as Russell thinks 47, for ‘exist’ has different meanings in them, as
we bring to the light by spelling out their ontological commitments, respectively as

(54) The concept of the existent round square involves the property of existence,

and

(55) The concept of the existent round square is empty.

(6) The dagger is over my head (said by one who is having a hallucination)

Like (1), (2), (3), even (6) may have different readings: (6 1), making an implicit or explicit
existential claim; (65), carrying an existential presupposition; and (64), concerning a merely
intentional object (i.e., merely describing a subjective perception). Readings of the first kind are
correctly explained by Russell, those of the second kind by Frege and Strawson, while for the third
kind the right answer is Meinong’s; but it must be stressed that Meinong’s doctrine, in our
interpretation, does not contrast with Russell’s explanation of the first kind and Frege-Strawson’s
explanation of the second kind.
Actually, in this case there are further possible readings: (6 2) and (63), where the hallucination
concerns the dagger’s spatial position, not its existence; but this does not represent a problem for
anyone, as the dagger is a real one, and the sentence is not empty (we are supposedly describing a
real hallucination, not a fictional one like Macbeth’). Now,

(61) There exists exactly one thing that is a dagger [and salient in the present context], and it is over
my head

is false, and it is rendered as such by Russell (obviously, for this just is Russell’s chosen reading of
(6)). But one could also comment, in Meinong’s style, that (6 1) speaks of a nonexistent object (that
one thing that is both a dagger and above my head), and it is false because the object designated (not
referred to) by its subject does not have the properties (existence and spatial position) predicated by
its predicates. No contradiction arises, for speaking of X does not imply that X exists, as we noticed.
As for

(65) The dagger [whose existence I presuppose, without asserting it] is over my head,

it is neither true nor false, because of its false presupposition, as explained by Frege and Strawson;
from Meinong’s point of view, rather than using the ambiguous explanation that it lacks truth-value
because ‘the dagger’ refers to an incomplete object, we can adopt Strawson’s explanation and gloss
it as follows: it would not be possible to understand (6) as de re and presupposing a false existential
claim about the dagger, as in (65), unless one could speak or think of nonexistent objects, i.e.
designate them, or dwell upon their concept; but since this is possible, the speaker can understand
47
Russell (1905a), (1905b).
(6) as de re; so, he or she does presuppose a false existential claim, and this frustrates his or her
attempt to say something that is either true or false. Finally,

(64) The dagger [the one I am seeing, whether real or not] is over my head [i.e., that is precisely the
position where I see that dagger as being].

The truth of sentences like this (and (2), (3), (4)) is precisely what Meinong wanted his doctrine to
explain, and it is not satisfactorily explained either by Frege, Russell, or the neo-Russellian
approach: the speaker has the visual experience of a nonexistent object, and speaks of that object,
without asserting or implying anything concerning its existence. If we wish we may say that (6 4) is
true because the perceived object has the perceived property, but we are not actually committing
ourselves to the existence of such an object: (64) is just another way to describe a subjective
experience, and it is true if the experience is as (64) describes it.
Even in this case, the Russellian and neo-Russellian intuition that the truth-value of sentences
of this form is determined by representations (concepts, meanings, or, in this case, visual images),
rather than by things, is correct; but we must forbear from placing such representations as the direct
object of the original attitude (thinking of, or, in this case, seeing), a strategy that would render (6 4)
as

(66) I am seeing a visual image of a dagger, and I see it as above my head.

This would be a categorial mistake, for one does not see visual images, but sees objects by having
visual images of them. Just as thinking of X is dwelling upon the concept of X, seeing X (in the
subjective meaning of ‘seeing’) is having a visual image of X. The mistake is caused by mixing up
two different accounts, each correct in its own setting: the introspective observation sentence “I am
seeing a dagger”, and the sentence of psychological theory “I am having a visual image of a
dagger”. The relation between these latter two sentences is the same as that between (4) and (4 6), and
it is analogous to the relation between the everyday language sentences (2) and (3) (understood de
dicto) and their theoretical explications (212) and (313).
The above categorial mistake is eliminated when (66) is turned into

(67) There exists exactly one visual image, such that I am now having it, and is an image of a dagger
over my head,
or

(68) I see that there exists exactly one thing that is a dagger and is over my head,

where ‘to see [that]’ is a propositional verb and has a subjective meaning like in (6 4): i.e., just like
‘to think [that]’, it does not imply that its subordinate proposition is true. In § 2 and § 5 we noticed
that Russell says that we never need to assume the existence of intentional objects, but only of
thoughts, concepts, pictures or verbal descriptions, i.e., subjective representations of one sort or
another. So, it might be suggested, (67) and (68) are possible renderings of (64) in Russell’s
quantificational analysis. However, neither (67) nor (68) have the same truth-conditions as (64): for,
suppose I actually do not see anything, but just pretend, with my gestures and the expression of my
face, that I am having this hallucination. Then, if I utter (6 4), what I say is neither true nor false,
while if I utter (67) or (68) what I say is false. Once again, Russell’s idea that every definite
description conceals an existential claim (as opposed to an existential presupposition) turns out to be
wrong. A logical equivalent of (64), instead, is

(69) The visual image [which I am presupposing (but not asserting) I am having] is that of a dagger
over my head.

Still, as by now we know, (6 9) cannot supersede (64), because it does not have the same meaning.
Meinong’s “theory of objects” may be interpreted as a way to stress both the ineliminability of such
locutions as (64), (2), (3) and (4), and their harmlessness: there is nothing wrong with describing
ourselves as speaking of, or thinking of, or seeing, nonexistent objects.

Mario Alai
Università di Urbino
[email protected]

References

Alai, Mario (1994), Modi di conoscere il mondo, Milano, Franco Angeli 1994.
Barbero, Carola, and Raspa, Venanzio (eds.) (2005), Il pregiudizio a favore del reale. La teoria
dell’oggetto di Alexius Meinong fra ontologia e epistemologia, Rivista di Estetica [n.s. Vol. 30/3],
2005.
Carnap, Rudolf (1947), Meaning and Necessity, Chicago, The University of Chicago Press 1947.
Chisholm, Roderick M. (1982), “Beyond Being and Nonbeing”, in Brentano and Meinong Studies,
Amsterdam, Rodopi 1982, pp. 53-67.
Fodor, Jerry (1975), The Language of Thought, New York, Crowell 1975.
Frege, Gottlob (1891), “Funktion und Begriff”, Vortrag, gehalten in der Sitzung von 9 Januar 1981
der Jenaischen Gesellshaft für Medizin und Naturwissenschaft, Jena, Herman Pohle
– (1892), “Sinn und Bedeutung”, Zeitschrift für Philosdophie und philosophische Kritik, 100 (1892),
pp. 25-50.
Husserl, Edmund (1894), “Vorstellung und Gegenstand”, hrsg. von K. Schuhmann, Brentano
Studien, 3 (1990/1991), pp. 137-176.
– (1901), “Draft of a letter to Marty”, in E. Husserl, Briefwechsel, vol. I: Die Brentanoschule, hrsg.
von K. Schuhmann, Dordrecht/Boston/London, Kluwer 1994, pp. 75-83.
– (1901/13), Logische Untersuchungen. Zweiter Band: Untersuchungen zur Phänomenologie und
Theorie der Erkenntnis, hrsg. von U. Panzer (Husserliana XIX), Dordrecht, Kluwer, 1984.
– (1913), Ideen zu einer reinen Phänomenologie und phänomenologischen Philosophie. Erstes
Buch: Allgemeine Einführung in die reine Phänomenologie, hrsg. von K. Schuhmann (Husserliana
III), Den Haag, Nijhoff, 1976.
Kripke, Saul (1972), “Naming and Necesity”, in D. Davidson and G. Harman (eds.) Semantics of
Natural Language, Dordrecht, Reidel 1972, pp. 253-355, 762-769.
Lewis, Clarence Irving (1943-1944), “The Modes of Meaning”, Philosophy and Phenomenological
Research, 4 (1943-1944), pp.236-250.
Meinong, Alexius (1902), Über Annahmen, Leipzig, Barth, 1902.
– (1904) “Über Gegenstandstheorie”, in Untersuchungen zur Gegendstandstheorie und
Psychologie, hrsg. von A. Meinong, Leipzig, Barth, 1904.
Napoli, Ernesto (2000), “Finti nomi”, in G. Usberti (ed.) Modi dell’oggettività, Milano, Bompiani,
2000, pp.197-221.
Orilia, Francesco (2002), Ulisse, il quadrato rotondo e l’attuale re di Francia, Pisa, ETS, 2002.
– (2005), “La libertà di assunzione nella filosofia analitica contemporanea”, in Barbero and Raspa
(eds.) (2005), pp. 91-109.
Poli, Roberto (2005), “Meinong, filosofo empirico”, in Barbero and Raspa (eds.) (2005), pp. 116-139.
Putnam, Hilary (1973), “Meaning and Reference”, Journal of Philosophy 70 (1973), pp. 699-711.
– (1975) “The Meaning of ‘Meaning’”, in H. Putnam, Mind, Language and Reality, Cambridge,
University Press 1975, pp. 215-271.
– (1981), Reason, Truth and History, Cambridge-New York-Melbourne, Cambridge University Press
1981.
Quine, Willard van Orman (1960), Word and Object, Cambridge, Mass., Massachussetts Institute of
Technology 1960
Raspa, Venanzio (2005), “Forme del più e del meno in Meinong”, in Barbero and Raspa (eds.)
(2005), pp. 185-219.
Russell, Bertrand (1905a), “On Denoting”, Mind, 14 (1905), pp. 479-493.
– (1905b), Review of: A. Meinong, Untersuchungen zur Gegenstandstheorie und Psychologie,
Mind, 14 (1905), pp. 530-538.
– (1919) Introduction to Mathematical Philosophy, London, Allen and Unwin, 1919.
Searle, John (1958), “Proper Names”, Mind 67 (1958), pp. 166-173.
Strawson, Peter (1950), “On Referring”, Mind 59 (1950), pp. 320-344.
Wittgenstein, Ludwig (1953), Philosophische Untersuchungen, Oxford, Blackwell 1953.

List of examples

(1) The present king of France is bald.

(11) There exists a present king of France, and he is bald.
(12) There exists exactly one thing that is presently king of France, and it is bald.
(13) The concept [THE[PRESENT KING OF FRANCE]] implies the property of being bald.
(14) Being bald is a property F such that there exists exactly one thing that is both presently king of
France and F.
5
(1 ) The concept [THE[PRESENT KING OF FRANCE]] is bald.
(16) The one thing [whose existence is here presupposed, but not asserted] that is presently king of
France is bald.
7
(1 ) The present king of France as such is bald.
(18) The concept of the present king of France involves the property of being bald.

(2) The golden mountain is golden.

(21) There exists exactly one thing that is both a mountain and golden, and it is golden.
(22) [Implicit: there exists a golden mountain, and] the [aforementioned] golden mountain is
[obviously] golden.
3
(2 ) Every golden mountain is golden.
(24) For any object x, if x is a mountain and golden, then x is golden.
(25) For any object x, if x is a mountain and golden, then x is not golden.
(26) If the golden mountain existed, it would be golden.
(27) It is necessary that a golden mountain is golden.
(28) The concept [THE[GOLDEN & MOUNTAIN]] involves the property of being golden.
(29) Being golden is a property F such that there exists exactly one thing that is a mountain, golden,
and F.
10
(2 ) The concept [THE[GOLDEN & MOUNTAIN]] is golden.
(211) The golden mountain as such is golden
(212) The concept of the golden mountain involves the property of being golden.

(3) Sherlock Holmes was clever.

(31) There existed exactly one thing that was an English detective, lived at 221b Baker Street, etc.,
and it was clever.
2
(3 ) Sherlock Holmes [implicit: existed, and he] was clever.
(33) If Sherlock Holmes had existed he would have been clever.
(34) Conan Doyle writes (or implies) that Sherlock Holmes was clever.
(35) Conan Doyle writes (or implies): “Sherlock Holmes is clever”.
(36) Being clever is a property F such that there exists exactly one thing that is an English detective,
lives at 221b Baker Street, etc., and is F.
(3 ) [λF ∃1x (ED(x) & LB(x), …&F(x))](clever).
7

(38) [λF ∃1x (ED(x) & LB(x), …&F(x))].

(39)[THE[ED & LB & ….etc.]].
(310) The concept [THE[ENGLISH DETECTIVE & LIVING AT 221b BAKER STREET & …]]
implies the property of being clever.
11
(3 ) Sherlock Holmes had a cold on March, 3rd, 1891.
(312) The concept [THE[ENGLISH DETECTIVE & LIVING AT 221b BAKER STREET & …]]
implies the property of having a cold on March, 3rd, 1891.
13
(3 ) The concept of the English detective living at 221b Baker Street, etc., involves the property of
being clever.
14
(3 ) The one thing [whose existence here is not asserted, but presupposed] that is an English
detective, living at 221b Baker Street, etc., had a cold on March, 3rd, 1891.
15
(3 ) Conan Doyle writes (o implies) that Sherlock Holmes had a cold on March, 3rd, 1891.
(316) The concept of the one English detective living at 221b Baker Street, etc., implies the property
of having a cold on March, 3rd, 1891.

(4) Frank thinks of the golden mountain.

(41) There exists exactly one thing which is both a mountain and golden, and Frank thinks of it.
(42) Frank thinks that there exists exactly one thing which is both a mountain and golden.
(43) Frank thinks “Golden mountain”.
(44) Frank thinks of the concept [THE[MOUNTAIN & GOLDEN]].
(45) There exists a concept involving the properties of being a mountain, and golden, and Franks
thinks of it.
6
(4 ) Frank dwells upon the concept of the golden mountain.

(5) The round square does not exist.

(51) There exists nothing that is both square and round.
(52) The existent round square exists.
(53) The existent round square does not exist.
(54) The concept of the existent round square involves the property of existence.
(55) The concept of the existent round square is empty.

(6) The dagger is over my head (said by one who is having a hallucination).
(61) There exists exactly one thing that is a dagger [and salient in the present context], and it is over
my head.
2
(6 ) The dagger [i.e.: my friend’s dagger] is over my head.
(63) There exists exactly one thing that is a dagger and belongs to my friend, and it is over my head.
(64) The dagger [the one I am seeing, whether real or not] is over my head [i.e., that is precisely the
position where I perceive that dagger as being].
5
(6 ) The dagger [whose existence I presuppose, without asserting it] is over my head.
(66) I am seeing a visual image of a dagger, and I see it as above my head.
(67) There exists exactly one visual image, such that I am now having it, and is an image of a dagger
aver my head.
8
(6 ) I see that there exists exactly one thing that is a dagger and is over my head.
(69) The visual image [which I am presupposing (but not asserting) I am having] is that of a dagger
over my head.