Balaguer 1998

International Phenomenological Society
Philosophy and Phenomenological Research
Attitudes without Propositions

Author(s): Mark Balaguer
Source: Philosophy and Phenomenological Research, Vol. 58, No. 4 (Dec., 1998), pp. 805-826
Published by: International Phenomenological Society
Stable URL: http://www.jstor.org/stable/2653723
Accessed: 01-11-2015 23:41 UTC
REFERENCES
Linked references are available on JSTOR for this article:
http://www.jstor.org/stable/2653723?seq=1&cid=pdf-reference#references_tab_contents
You may need to log in to JSTOR to access the linked references.
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at http://www.jstor.org/page/
info/about/policies/terms.jsp
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.
Wiley, International Phenomenological Society and Philosophy and Phenomenological Research are collaborating with
JSTOR to digitize, preserve and extend access to Philosophy and Phenomenological Research.
http://www.jstor.org
This content downloaded from 134.184.26.108 on Sun, 01 Nov 2015 23:41:36 UTC
All use subject to JSTOR Terms and Conditions
Philosophy and Phenomenological Research
Vol. LVIII, No. 4, December 1998
AttitudesWithoutPropositions1
MARK BALAGUER
California State University, Los Angeles
This paper develops a novel version of anti-platonism,called semanticfictionalism.The

view is a response to the platonistargumentthat we need to countenancepropositionsto
account for the truth of sentences containing 'that'-clause singular terms, e.g., sen-
tences of the form 'x believes that p' and 'a means that p'. Briefly, the view is that (a)
platonists are right that 'that'-clauses purportto refer to propositions, but (b) there are
no such things as propositions, and hence, (c) 'that'-clause-containingsentences of the
above sort are not true-they are useful fictions. Semantic fictionalismis an extension of
HartryField's mathematicalfictionalism,but my defense of the view is not analogous to
his. One of the many virtues of my defense is its generality: it explains how we can
adopt a fictionalist stance towardsall abstractsingularterms, e.g., mathematicalsingular
terms and 'that'-clauses.
1. INTRODUCTION
Most of the traditional arguments for the existence of abstract objects (i.e.,
non-physical, non-mental, non-spatio-temporal objects) can be understood as
inferences to the best explanation. Platonists argue that some phenomenon
(e.g., resemblance) can only be explained by appealing to abstract objects of
some sort (e.g., universals) and conclude from this that there must actually be
some abstract objects. One such argument holds that we need to countenance
propositions2 to account for the truth of various sentences containing 'that'-
clauses, e.g., sentences of the form 'x believes that p'. (Arguments of this
kind can be traced back at least to Frege, but the version I will describe here
is most closely related to the versions developed recently by George Bealer
This paper was read at the City University of New York GraduateCenter in April, 1996,
where I received several helpful comments from various members of the audience, most
notably JerroldKatz. I also received very helpful written comments on an earlier draft of
the paper from Seth Crook, Russell Dale, HartryField, Bob Hanna, and Adam Vinueza. I
would like to thankall of these people.
2 There are various views of the nature of propositions, but it won't matter here which of
these views is correct,because the only featureof propositionsthat will be relevant to my
argument is their abstractness, and this is something that all the standard views agree
upon.
ATTITUDESWITHOUTPROPOSITIONS 805
and Stephen Schiffer.3)In this paper, I will respond to all argumentsof this
general kind. I will do this not by arguingthat we can account for the fact in
question-i.e., the fact thatthereare true 'that'-clause-containingsentences-
without appealing to propositions, but rather, by arguing that we have no
good reason to suppose that there really is a fact here at all. Thus, I will be
introducing what I think is a fairly novel version of anti-platonism, one
which (a) admits that platonists are right about the truthconditions of 'that'-
clause-containing sentences (i.e., admits that such sentences really are
"about"4propositions) but nonetheless, (b) maintainsthat there are no such
things as propositions, and thus, (c) concludes that the truth-conditionsof
'that'-clause-containingsentences are never satisfied-i.e., that there are no
true 'that'-clause-containingsentences (except for those which are vacuously
true). I will call this view semanticfictionalism.
Now, prima facie, semantic fictionalism might seem wildly implausible
or even downrightcrazy. But we can dispense with this primafacie worryand
lay bare a certain attractiveness to semantic fictionalism by bringing out
some of the parallels between it and Hartry Field's mathematical
fictionalism.5 I take this up in section 2. I also lay out the Frege-Bealer-
Schiffer argumentthere and say a few words aboutmy response to that argu-
ment before giving the meat of the response in section 3. The stance I adopt
in sections 2 and 3 requiresme to reply to an argumentthat is closely related
to the Frege-Bealer-Schiffer argument, viz., the argumentthat we need to
countenance propositions to account for the meaningfulness of sentences.
(This argumentalso goes back at least to Frege; a contemporaryadvocate is
Jerrold Katz.6)I respond to this argumentin section 4 by merely extending
the position developed in sections 2 and 3.
See Gottlob Frege, "UeberSinn und Bedeutung,"translatedby H. Feigl as "On Sense and
Nominatum"in Philosophy of Language, ed. A.P. Martinich(Oxford: Oxford University
Press, 1990); and George Bealer, "Universals,"Journal of Philosophy, 90 (1993): 5-32;
and Stephen Schiffer, "A Paradox of Meaning," Nous, 28 (1994): 279-324. Chapters 4
and 5 of Schiffer's Remnants of Meaning (Cambridge,Massachusetts:MIT Press, 1987)
are also relevant here, althoughhis overall point there is different.
On the use of the term 'about' that I adopt in this paper, the claim that a sentence cais
about an object x does not entail that x exists. Thus, for instance, 'Oliver Twist was a
boy' is about Oliver Twist, even though there was never any such person. In order to
remind the readerthat I am using 'about' in this way, I will try to put scare quotes around
that word whenever I say that a sentence is about an object whose existence is in ques-
tion.
See HartryField, Science WithoutNumbers(Princeton:PrincetonUniversity Press, 1980).
See JerroldKatz, "CommonSense in Semantics,"in New Directions in Semantics, ed. E.
LePore (London: Academic Press, 1987).
806 MARK BALAGUER
2. MATHEMATICAL FICTIONALISM AND SEMANTIC
FICTIONALISM
There is a famous Fregean argument for mathematical platonism that pro-
ceeds by arguing that we need to countenance abstract mathematical objects to
account for the truth of mathematical sentences.7 For example, it is claimed if
we grant that
(1) 3 is prime
is true, then we also have to grant that there is such a thing as the number 3
(and that it is an abstract object). The best response to this argument, in my
opinion, is the one given by mathematical fictionalists. The idea here is to
grant the central claim of the Fregean argument-that if sentences like (1) are
true, then there are abstract mathematical objects and platonism is true-but
to maintain that sentences like (1) are simply not true. The reason such
sentences are not true, according to fictionalism, is that there are no such
things as mathematical objects. In other words, the problem is that mathe-
matical singular terms, e.g., '3', are vacuous, i.e., fail to refer. Thus, (1) is
false (or not true8) for the same reason that 'Oliver Twist lived in London' is
false-because just as there was never any such person as Oliver Twist, so
there is no such thing as 3.9
(One might worry that by taking this line, we lose the distinction between
mathematically sound sentences like (1) and mathematically unsound sen-
tences like '4 is prime'. But as Field has shown, we can block this worry by
merely noting that '3 is prime' is true-in-the-story-of-mathematics, whereas
'4 is prime' is not-just as 'Oliver lived in London' is true-in-the-story-of-
Oliver-Twist, whereas 'Oliver lived in Paris' is not.'0 The real worry about
See Gottlob Frege, Der Grundlagen die Arithmetic, translatedby J. L. Austin as The
Foundationsof Arithmetic(Oxford:Basil Blackwell, 1953).
8 That sentences like (1) are false is not essential to mathematical fictionalism. What is
essential to that view is that (a) there are no mathematicalobjects, and hence, (b) math-
ematical singular terms are vacuous. Whether this means that sentences like (1) are
false, or lacking a truth value, or something else, depends upon our theory of vacuity. I
will adoptthe view that such sentences arefalse, but nothing importantwill turnon this.
9 Fictionalists allow that some mathematical sentences-e.g., 'If 3 is prime, then 3 is
prime'-are true, because they are vacuously true. I will ignore this complication and
speak as if fictionalists believe that all mathematicalsentences are false.
10 What does it mean to say: "The sentence ca is true-in-the-story-2:"?The best and most
straightforwardinterpretationis to take this as saying that a certain sentence type (viz.,
cY)is a member of a certainset of sentence types (viz., l). Now, given this, it might seem
that the appeal to the predicate '...is-true-in-the-story-of-mathematics'has not provided
fictionalistswith a way of distinguishing'3 is prime' from '4 is prime'. For since sentence
types and sets of sentence types are abstract objects, it seems that fictionalists have to
maintainthat
"'3 is prime' is true-in-the-story-of-mathematics"
and
"'4 is prime' is true-in-the-story-of-mathematics"
ATTITUDES WITHOUT PROPOSITIONS 807
fictionalism is that it seems incompatible with the applicability of mathe-
matics. I will returnto this shortly.)
Semanticfictionalismcan be developed in an exactly analogousway. Cor-
responding to the Fregean argumentthat we need to endorse platonism to
account for the truthof mathematical sentences is the Frege-Bealer-Schiffer
argumentthat we need to endorseplatonismto accountfor the truthof 'that'-
clause-containing sentences (or as I will call them from now on, 'that'-sen-
tences). The Frege-Bealer-Schifferargumentproceeds in two steps: first, it is
argued that 'that'-clauses are referential singular terms, and second, it is
argued that the only things that could be the referents of 'that'-clauses are
propositions. The first step is motivated by appealing to certain valid argu-
ments, e.g.,
Jenny believes thatIdgie is a dog

Therefore,Jennybelieves something
Platonists claim that the only way to account for the validity of this
argumentis to treatit as an existential generalization,i.e., to take the logical
form of the argumentto be:
B(j, the referentof 'thatIdgie is a dog')

.,. (3x)B(j, x)
where 'j' denotes Jenny and 'B' expresses a two-place belief relation. But if
this is right, then it seems that we have no choice but to conclude that 'that
Idgie is a dog' is a referentialsingularterm.The second premise of the Frege-
Bealer-Schifferargument-that the only things that could be the referentsof
'that'-clauses are propositions-is motivated simply by ruling out all of the
competitors, i.e., by arguing that things like facts, public-language
sentences, and private-languagesentences couldn't be the referentsof 'that'-
clauses. Several different reasons have been given, by various philosophers,
for concluding that such things could not be the referentsof 'that'-clauses;I
are on all fours. (In particular,they have to maintainthat both of these sentences are, like
purely mathematicalsentences, fictional-i.e., strictly speaking false.) We can solve this
problemby turningour attentionto certainconcrete objects that fictionalistsdo believe in,
namely, tokens of '3 is prime' and '4 is prime'. Fictionalists maintainthat tokens of '3 is
prime' have a certain propertythat tokens of '4 is prime' do not have. We can describe
this propertyin platonistic terms as the propertyof being a token of a type that is true-in-
the-story-of-mathematics.But it seems that fictionalistsneed a nominalistic descriptionof
this property. The problem here is a special case of the general problem of the applica-
bility of mathematics (and other abstract-objecttalk), for the problem there is precisely
that there are features of the physical world that seem describable in platonistic terms
only. I will explain how fictionalistscan solve this generalproblemin section 3, and it will
be obvious that the special case under discussion here can be solved in the same way.
808 MARKBALAGUER
will not discuss any of these reasonshere, because I am going to concede this
point to platonists.
Now, just as anti-platonists can respond to the Fregean argument for
mathematicalplatonism by endorsingmathematicalfictionalism, so too, they
can respond to the Frege-Bealer-Schiffer argumentby endorsing semantic
fictionalism. The idea here is to grant the two centralpremises of the Frege-
Bealer-Schifferargument-that 'that'-clausesarereferentialsingulartermsand
that the only plausible suggestion about what they might refer to is the
platonistic suggestion that they refer to propositions-but to maintain that
since there are, in fact, no such things as propositions, 'that'-clausesare vac-
uous terms. In other words, semantic fictionalists admit that 'that'-clauses
purport to refer to propositions,but they maintainthat, in point of fact, they
do not refer at all. And so just as mathematicalfictionalists hold that mathe-
matical sentences like (1) are false (or not true), semantic fictionalists hold
that 'that'-sentenceslike
(2) That 0. J. is a bachelorentails that 0. J. is unmarried
are false (or not true).

(The parallelbetween semanticfictionalismand mathematicalfictionalism
is really quite deep. For instance, like mathematicalfictionalists, semantic
fictionalists acknowledge that some 'that'-sentences--e.g., 'If I believe that
birds fly, then I believe that birds fly'-are vacuously true (although,again, I
will usually ignore this complicationand speak as if they hold that all 'that'-
sentences are false). And semantic fictionalists distinguish sentences like (2)
from sentences like 'That 0. J. is a bachelorentails that 0. J. is a murderer'
by pointing out that the former, but not the latter, are true-in-the-story-of-
propositions.)
It seems to me, then, that by endorsingsemantic fictionalism, anti-platon-
ists succeed in putting the ball back into the platonist court. The question
now is what argument there is for the truthof 'that'-sentenceslike (2). The
central thesis that I will try to motivate in this paper is thatplatonists do not
have any good argumenthere at all.
Before beginning my argumentfor this thesis, I would like to make two
points. First, one might think that platonists don't need an argument here,
that it is simply obvious that 'that'-sentences like (2)-and mathematical
sentences like (1)-are true. This, I think, is just wrong. If the two central
premisesof the Frege-Bealer-Schifferargumentare correct-and platonistsand
fictionalists both think that they are-then sentences like (1) and (2) could
only be true if there exist abstractobjects. But the question of whetherthere
exist abstractobjects is controversialand non-obvious. Thus, (1) and (2) are
ATTITUDESWITHOUTPROPOSITIONS 809
also controversial and non-obvious." The second point I want to make is that
in establishing that platonists have no argument for the truth of 'that'-sen-
tences like (2), I will not be arguing that semantic fictionalism is true, or
even that it's superior to platonism. I will merely be defending it against a
certain attack (or as I put it above, blocking a certain argument for platon-
ism). In the end, I do not favor semantic fictionalism over platonism, because
I do not think there are any good arguments for the falsity of 'that'-sentences.
Thus, what I think is that the Frege-Bealer-Schiffer argument shows that
either platonism or semantic fictionalism is true but that we don't have any
good argument that tells us which one is true.
In any event, I begin my discussion by asking how platonists might try
to argue that some (non-vacuous) 'that'-sentences are true. The best platonist
strategy here-indeed, the only promising strategy-is to construct an argu-
ment analogous to the one they have used in the mathematical case, i.e., the
argument they have used to motivate the claim that mathematical sentences
like (1) are true. The argument I have in mind is the Quine-Putnam indis-
pensability argument,'2 which holds (in a nutshell) that we have to allow that
at least some mathematical sentences and theories are true, because they are
indispensable parts of certain empirical theories that we believe to be true.
This argument can be used to motivate the truth of various 'that'-sentences as
well as mathematical sentences. For while 'that'-clause singular terms are not
used as frequently in empirical science as mathematical singular terms are,
they do appear is some of our empirical theories, most notably belief psy-
chology, which uses sentences like
(3) Floyd believes that Bill Clinton is president
to explain human behavior.

(Some people doubt that folk concepts like 'believes' belong in a mature
scientific psychology, but this issue is entirely irrelevant here. It doesn't
matter whether sentences like (3) belong in scientific theory. All that matters
is that such sentences seem to express facts about the belief states of people
like Floyd and, indeed, that they seem indispensable to the project of
characterizing these facts. Whether they are indispensable to scientific theory
doesn't really matter, although I will couch my discussion in those terms.)
Something else which semantic fictionalists reject-but which might have seemed obvi-
ous, pre-theoretically-is the inference from 'Fa' to 'That Fa is true'. But again, the
invalidity of this inference is forced on us by the platonist's own argument;for the Frege-
Bealer-Schiffer argumentshows that whereas 'Fa' is about a, 'That Fa is true' is about
the propositionthatFa.
12 See the last section of W. V. 0. Quine, "Two Dogmas of Empiricism,"reprintedin From
a Logical Point of View (New York: Harper and Row, 1961); and chapters V-VIII of
Hillary Putnam,The Philosophy of Logic (New York: Harperand Row, 1971).
810 MARK BALAGUER
How can fictionalists respond to this indispensability argument?Well,
Field has responded to the version of the argument that's directed against
mathematical fictionalism by arguing that our mathematicaltheories are, in
fact, not indispensableto empirical science. He maintainsthat our empirical
theories can be nominalized-i.e., reformulatedin a way that (a) makes no
referenceto, or quantificationover, mathematicalobjects and (b) is still theo-
retically attractive-and that we should only endorsethe truthof the nominal-
ized versions of these theories. There have been a number of objections to
Field's argument,however, and it is not at all clear that it succeeds.'3 More-
over, even if it succeeds in the mathematicalcase, it is difficult to see how it
could be generalized, so that it would apply to the case of 'that'-sentencesas
well as to mathematics. But none of this matters, because there is another
strategy that fictionalists can employ here: they can admit that our empirical
theories make indispensableuse of mathematicsand simply account for this
in fictionalist terms. I have provided such an explanationelsewhere.'4In the
next section, I will show that this explanationcan be generalized,so that it is
an account not just of the use that physics makes of mathematical objects,
but also the use that belief psychology makes of propositions,and indeed, the
use that empiricalscience makes of abstractobjects.
3. A FICTIONALIST ACCOUNT OF BELIEF

I need to provide a fictionalistaccountof the (dispensableand indispensable)
applicationsof our platonistic theories (e.g., the "theoryof propositions")to
our empiricaltheories (e.g., the "theoryof belief psychology"). I will assume
here that in doing this, I need to maintainsome sort of realism about empiri-
cal science. That is, I will assume that it is not acceptable for fictionalists to
claim that the reason fictional platonistictheories are applicableto empirical
science is that empirical science is also fictional (i.e., that this applicability
should seem no more surprisingthan does the applicabilityof "RamboII" to
"RamboIII").I cannot discuss why I think this response is unacceptable,but
in a nutshell, the reason is that I think full-blown scientific anti-realismis an
untenable view. That is, I think that our scientific theories do express truths
about the physical world. Thus, the challenge facing fictionalists is to locate
a version of scientific realism that does not simply assert the truth of our
empirical theories (because that version of realism commits to the existence
of abstracta). The challenge is to explain how we can maintain that our
13 Most of these objections are discussed by David Malamentin his review of Field's book
in The Journal of Philosophy, 79 (1982): 523-34. I block one of these objections-viz.,
the objection that Field's method cannot be extended to cover quantum mechanics-in
my "Towardsa Nominalizationof QuantumMechanics,"Mind, 105 (1996): 209-26.
14 See chapterVII of my Platonism and Anti-Platonismin Mathematics(New York: Oxford
University Press, 1998); or see my "A Fictionalist Account of the Indispensable Appli-
cations of Mathematics,"Philosophical Studies, 83 (1996): 291-314.
ATrTrUDES WITHOUT PROPOSITIONS 811
empirical theories are strictly speaking false without committing to the
implausibleclaim thatthere are no truths"buried"in these theories.
My proposal is that fictionalists can endorse what I will call nominalistic
scientific realism, the view that the nominalistic content of empirical
science-i.e., what empirical science entails about the physical world-is
true (or mostly true-there may be some mistakes scatteredthroughit), while
its platonistic content-i.e., what it entails "about" a platonic realm of
abstractobjects-is false. The reason this view is a genuine form of scientific
realism is that it endorsesthe "completepicture"that empiricalscience paints
of the physical world, including the parts about so-called "theoreticalenti-
ties", e.g., electrons.
There is an immediateworry aboutnominalisticscientific realism that can
be expressed in two different ways. First, one might claim that if everything
empirical science says about the platonic realm is fictional, then much of
what it says about the physical world will come out false, and hence, even if
we preserve some of what empirical science says about the physical world,
we will not preserveit all, i.e., we will not preservethe completepicture that
empirical science paints of the physical world. Second, one might claim that
if everything empirical science entails about the physical world is true, then
what it entails about the platonic realm must also be true. Both claims arise
from the single worry that it is not possible to separate the nominalisticcon-
tent of empirical science from its platonistic content. I will respond to both
sides of this worry by arguing that this is possible; more specifically, I will
arguethat
(NC) Empiricalscience has a purely nominalisticcontent that capturesits

''completepicture"of the physical world,
and
(COH) It is coherent and sensible to maintainthat the nominalistic con-

tent of empirical science is true and the platonistic content of
empiricalscience is fictional.
(It might seem that this stance commits me to the claim that empirical
science can be nominalized,but we will see that it does not.) In any event, by
arguing for (NC) and (COH), I will essentially be arguing that nominalistic
scientific realism is a coherent view. Now, I will also argue that it's a plau-
sible view, but I don't need to show that it's true, because I'm not trying to
show thatfictionalism is true. All I want to show is that we have no good
reason to reject these two views.
My argument for (NC) and (COH) is based upon the obvious fact that
abstractobjects (if there are such things) are not causally relatedto anything
in the physical world. Notice, first, that if we assume that (NC) is true, then
812 MARK BALAGUER
the causal inertness of abstract objects suggests that (COH) is also true,
because it suggests that the truthvalue of the platonistic content of empirical
science is simply irrelevantto the truthvalue of its nominalistic content. We
can think of it this way: if all the abstractobjects in platonic heaven suddenly
disappeared, nothing would change in the physical world; thus, if empirical
science is true right now, then its nominalistic content would remain true,
even if the platonic realm disappeared;but this suggests that if there never
existed any abstractobjects to begin with, the nominalistic content of empir-
ical science could nonethelessbe true.
But the main point that needs to be made here is that the appeal to causal
isolation lends support to (NC) as well as to (COH). To put the argument
very quickly, it is this. Empirical science knows, so to speak, that abstract
objects are causally inert. That is, it does not assign a causal role to any
abstractentities. Thus, it seems that empiricalscience predicts that the behav-
ior of the physical world is not dependentin any way upon the existence of
abstractobjects. And this suggests that what empiricalscience says aboutthe
physical world-i.e., its complete picture of the physical world-could be
true even if there aren't any abstractobjects. That is, it suggests that (NC)
and (COH) are both true.
Now, as a segue into a more complete and adequate statement of the
argument,consider the following objection. "You seem to be assuming that
because empiricalscience doesn't ascribeany causal role to abstractobjects, it
doesn't ascribe any role to them at all. But this is wrong: in giving its pic-
ture of the physical world, part of what empirical science tells us is that cer-
tain physical systems are relatedin certainnon-causal ways to certainabstract
objects. Consider,for example, the sentence
(4) The physical system S is forty degrees Celsius.
You are quite right that in making this claim, we do not mean to assign any
causal role to the number 40-that we do not mean to suggest that the
number40 is responsible in any way for S's having the temperatureit has.
But nonetheless, we are saying something that involves the number40: we
are saying that S stands in a certain non-causal relation-viz., the Celsius
relation-to that number.Likewise, despite the fact that (3) does not imply
that the proposition that Clinton is president is causally responsible for
Floyd's believing this proposition, it does say that Floyd stands in a certain
non-causal relation-viz., the belief relation-to this proposition. Thus, it
seems that our empirical theories do not simply express some nominalistic
facts and some platonistic facts; rather,they express mixedfacts. And so it
seems that (NC) is false: empiricalscience does not have a nominalisticcon-
tent that capturesits complete pictureof the physical world."
ATlTUDES WITHOUT PROPOSITIONS 813
The person who objects in this way fails to appreciate the full significance
of the causal inertness of abstract objects. It is no doubt true that (4) says that
S stands in the Celsius relation to the number 40. But since 40 isn't causally
relevant to S's temperature, it follows that if (4) is true, it is true in virtue of
facts about S and 40 that are entirely independent of one another, i.e., that
hold or don't hold independently of one another. In other words, if we grant
that the number 40 isn't causally related to S-and this is beyond doubt-
then we are forced to say that while (4) does express a mixed fact, it does not
express a bottom-level mixed fact, i.e., that the mixed fact that (4) expresses
supervenes on more basic facts that are not mixed.15 In particular, it super-
venes on a purely physical fact about S and a purely platonistic fact about the
number 40. But this suggests that (4) has a nominalistic content that captures
its complete picture of S: that content is just that S holds up its end of the
"(4) bargain", i.e., that S does its part in making (4) true.16 (We might also
try to say that the nominalistic content of (4) is that the purely physical fact
behind (4)-i.e., the purely physical fact about S just mentioned-obtains.
But we have to be careful here. The purely physical fact behind (4) is a par-
ticular fact, presumably having something to do with kinetic energy. But the
nominalistic content of (4) is not that this particular fact holds; it couldn't
be, because (4) doesn't describe any such fact; e.g., it doesn't even broach the
topic of kinetic energy. Thus, all we can say here is that the nominalistic
content of (4) is that some purely physical fact that involves S holding up its
end of the "(4) bargain" obtains.)
The same goes for (3): it does say that Floyd stands in the belief relation
to the proposition that Clinton is president, but since this proposition isn't
causally relevant to Floyd's belief state, it follows that if (3) is true, it is true
in virtue of facts about Floyd and the proposition which are entirely indepen-
dent of one another. And, of course, the fact about Floyd on which (3) super-
venes is going to be a purely physical fact; in particular, it's going to be a
fact about his brain (or if externalism is true, his brain and his environ-
ment"7). Thus, it seems to follow that (3) has a nominalistic content that cap-
15 Strictly speaking, I should say: if (4) is true, then the mixed fact that it expresses super-
venes on more basic facts. For if (4) isn't true, then at least one of the facts in question
here won't really exist. I leave this proviso out for the sake of rhetoricalelegance, and in
what follows, I will do the same thing on a few differentoccasions.
16 One might worry that the sentence "S holds up its end of the '(4) bargain"'is not purely
nominalistic on the grounds that when we unpack the expression '(4) bargain', we will
encounter the platonistic lingo of (4), in particular,the term 'forty'. But fictionalists can
maintainthat in talking about the "(4) bargain",we are really just talking about a token of
the sentence (4), and so the expressions in (4)-e.g., 'forty'-are being mentioned and
not used.
17 Externalism is the view that the contents of our beliefs are determined not just by our
internal states but also by our physical and social environments. See Hillary Putnam,
"Meaning and Reference," The Journal of Philosophy, 70 (1973): 699-711; and Tyler
Burge, "Individualismand the Mental,"Midwest Studies, 4 (1979): 73-121.
814 MARK BALAGUER
tures its complete picture of Floyd: that content just says that Floyd holds up
his end of the "(3) bargain". (That this really captures the complete picture
that (3) paints of Floyd follows from the fact that (3) doesn't express any bot-
tom-level mixed facts; for it follows from this that all (3) tells us about
Floyd is that he holds up his end of the "(3) bargain".18)
(When I say that the fact about Floyd upon which (3) supervenes is a
purely physical fact, what I mean is that it is a purely nominalistic fact.
Thus, even if Cartesian dualism were true, my argument would still hold
water. For in that case, (3) would still supervene upon two entirely indepen-
dent facts, one a purely nominalistic fact about Floyd and the other a purely
platonistic fact about the proposition that Clinton is president. The only dif-
ference would be that the nominalistic fact would be (at least partially) men-
talistic. And this is also why my argument is consistent with externalism.
For even if the fact about Floyd upon which (3) supervenes is partially socio-
logical-concerning, say, the conventions prevailing in Floyd's linguistic
community-it is still nominalistic. The bottom line is this: even if mental-
istic or environmental factors partially determine what beliefs we have, it is
still true that causally inert propositions (if there are such things) do not
determine-even partially-what beliefs we have.)
It should be clear that considerations of the above sort will bring us to the
same conclusions with regard to all of empirical science. For since no
abstract objects are causally relevant to the physical world, it follows that
none of our mixed sentences express bottom-level mixed facts and, hence,
that empirical science has a nominalistic content that captures its complete
picture of the physical world-a nominalistic content that says that the phys-
ical world holds up its end of the "empirical-science bargain". Thus, I con-
clude that (NC) is true. To recapitulate, the argument for this rests on three
premises, viz.,
(i) Abstract objects-if there are such things-are not causally relevant
to the operation or state of the physical world;
(ii) If (i) is true, then the mixed facts expressed by empirical science
supervene on more basic facts that aren't mixed, i.e., that are either
purely nominalistic (i.e., purely physical, or neurological, or
whatever) or else purely platonistic; and
18
I suppose that one might claim that (3) tells us something else about Floyd, viz., that he
stands in the belief relationto the propositionthat Clinton is president.But this is just nit-
picking: the point is that when we move from (3) to its nominalistic content, we do not
lose any importantpart of our pictureof Floyd.
(iii) If the consequent of (ii) is true, then (NC) is true, i.e., empirical
science has a nominalistic content that captures its complete picture
of the physical world.
And if what I have argued is correct, then it seems that (COH) is also true,
i.e., that it is coherent to believe the nominalistic content of empirical
science while maintaining that its platonistic content is purely fictional.
This is all I really need to argue in order to block the Frege-Bealer-Schiffer
argument for platonism. But it is worth pointing out that nominalistic
scientific realism can be shown to be not just coherent but actually quite
plausible. Before I do this, however, I would like to address two different wor-
ries that people might have about the above argument.
First, one might be worried that a scientific anti-realist could use my
argument strategy to motivate a view that endorsed the "macro-level content"
of empirical science but rejected its "micro-level content". The worry, of
course, is that this view is so implausible, that it undermines my argument
strategy. The fact of the matter, though, is that my argument strategy cannot
be used to motivate this view. The reason is that micro-level entities are
causally related to macro-level entities. Indeed, if all the micro-level entities
in the world suddenly disappeared, all the macro-level entities would disappear
along with them. Moreover, empirical science predicts this, because part of
its picture of the macro-level of the world is that it is composed of micro-
level entities. Thus, empirical science simply doesn't have a purely macro-
level content that captures its complete picture of the macro-level of the
world. Thus, there is no viable view that endorses the macro-level content of
empirical science but not its micro-level content.
Second, one might be worried that the claim that empirical science has a
nominalistic content that captures its complete picture of the physical world
is much more controversial than I have allowed, because it is essentially
equivalent to the claim that empirical science can be nominalized. But this
worry is just misguided: the claim that empirical science has a nominalistic
content that captures its complete picture of the physical world is different
from (and much weaker than) the claim that empirical science can be nomi-
nalized. The easiest way to appreciate this is to notice that empirical theories
wear their nominalistic contents on their sleeves. The nominalistic content of
a theory T is just that the physical world holds up its end of the "T bargain",
i.e., does its part in making T true. Thus, while the claim that empirical
science can be nominalized is highly controversial, the claim that it has a
nominalistic content that captures its complete picture of the physical world
is entirely trivial. Indeed, it is no more controversial than the claim that
abstract objects (if there are such things) are causally inert.
These remarks suggest that empirical science could have a nominalistic
content that captured its complete picture of the physical world even if it
816 MARK BALAGUER
couldn't be nominalized. But it doesn't yet tell us why this is so. The reason
is this: even if abstract-object talk were indispensable to empirical science,
abstract objects (if there are such things) would still be causally inert, and so
the truth of empirical science (assuming that it is true) would still supervene
upon two independent sets of facts, viz., a set of purely physical (or more
accurately, nominalistic) facts and a set of purely platonistic facts. What
indispensability would imply is that we could never describe all of these
purely physical facts in an attractive nominalistic theory. But there would
still be such facts, and it would still be true that such facts could obtain even
if there were no such things as abstract objects. Thus, empirical science
would still have a nominalistic content that captured its complete picture of
the physical world-a nominalistic content that says that the physical world
holds up its end of the "empirical science bargain"-and it would still be
coherent and sensible to endorse this nominalistic content while maintaining
that the platonistic content of empirical science is fictional. In short, the
point here is that it doesn't matter whether our theories can be separated into
the purely nominalistic and the purely platonistic, because it already follows
from the causal inertness of abstract objects that the bottom-level facts are
separated in this way.
(These considerations provide a strong reason for preferring my response
to the Quine-Putnam argument over Field's. On my view, even if empirical
science cannot be nominalized, it is still reasonable to believe only its nomi-
nalistic content and to treat its platonistic content as fictional. Thus,
fictionalists do not have to replace our current scientific theories with nomi-
nalistic theories. Rather, they can accept the platonistic versions of our
empirical theories as they stand. The only thing they need to point out is that
when they "accept" these theories, they only commit to the truth of their
nominalistic contents.)
Let me turn now to the task of showing that nominalistic scientific real-
ism is not just coherent but actually very plausible. To understand why this
is so, we need merely to understand the role that abstract-object talk plays in
empirical science. We have seen that our empirical theories do not take
abstract objects to be causally relevant to the operation or state of the physi-
cal world. Why, then, do they contain any platonistic talk at all? The answer
is this:
(TA) Empirical theories refer to abstract objects (most notably mathemati-

cal objects, but also other kinds of abstract objects, e.g.,
propositions) in order to construct theoretical apparatuses (or
descriptive frameworks) in which to make assertions about the
physical world.
In other words, empirical theories do not make claims of the form: 'physical
(or biological, or psychological, or whatever) phenomenon x occurs because
the platonic realm has nature y'. Rather, they make claims of the form: 'the
behavior (or state) of physical (or biological, or psychological, or whatever)
system S can be understood in terms of the platonistic structure M as
follows:...' Thus, abstract-object talk appears in our empirical theories as a
mere descriptive aid: by speaking in terms of the real number line, or a
Hilbert space, or the network of propositions, we simply make it easier to
say what we want to say about the physical world.
(While it is true that abstract-object talk only appears in our empirical
theories as a descriptive aid, it would be an oversimplification to claim that
this is the only role that it plays in empirical science. The reason is that there
is more to science than the conjunction of its theories, and one of the things
that scientists do, in addition to stating theories, is make certain kinds of
inferences, and when they do this, they often use mathematics. But we can
ignore this complication here, because the points that I will make about the
descriptive role of abstract-object talk apply equally well to its inferential
role. 19)
It seems to me that (TA) is so obvious that it hardly requires argument.
But to say just a few words here, let us consider the two examples we have
been discussing, i.e., (3) and (4). It seems very clear that the only reason we
refer to the number 40 in (4) is that it provides a convenient way of saying
what S's temperature state is. More generally, the point is that the Celsius
scale correlates different temperature states with different numbers, so that the
numerals serve as names of the temperature states, or to use the more com-
mon lingo, the numbers represent the temperature states. (The reason it is
convenient to use numerals here, rather than ordinary names like 'Bob' and
19
In order to take the same line on the inferentialrole of abstract-objecttalk that I take on
its descriptive role, I do not need it to be the case that there are nominalistic versions of
our platonistically formulatedarguments.But I do need it to be the case that if we have a
(sound) argumentfor C that takes Pl,...,Pn as premises and that's formulatedin platonis-
tic terms, so that at least one member of {Pl,...,Pn}-and perhaps also C-refers to, or
quantifies over, abstractobjects, then whenever the nominalistic content of {Pl,...,Pn } is
true, the nominalisticcontent of C is also true. But I think it's prettyobvious that this is the
case. For if the given argumentis really sound, then whenever {Pl,...,Pn} is true, C is
also true. Thus, whenever {Pl,...,Pn} is true, the nominalistic content of C is also true,
since it is included in C. But it follows from this that wheneverthe nominalistic content of
{Pl,...,Pn} is true, the nominalistic content of C is also true (which, again, is just what I
need) because there is nothing in {Pl,...,Pn} but not in the nominalistic content of
{Pl,...,Pn} that's at all relevant to whether the nominalistic content of C is true. This is
simply because (a) what's in {P1,...,Pn} but not in the nominalistic content of {P1,...,Pn}
is just the platonistic content of (P1,...,Pn}, and this is solely about abstractobjects, i.e.,
not aboutthe physical world at all; and (b) the nominalisticcontent of C is solely about the
physical world and not about abstract objects at all. In short, the platonistic content of
{Pl,...,Pn} is not relevant to the truth value of the nominalistic content of C, because
abstractobjects (if there are such things) are causally inert.
818 MARK BALAGUER
'Ted', is that the various temperature states are related to one another in a way
that is analogous to the way in which the real numbers are related to one
another.) Likewise, the reason we refer, in (3), to the proposition that Clin-
ton is president is that this provides an easy way of expressing a certain belief
state of Floyd's. And more generally, the point is that just as the empirical
structure of temperature states can be represented by the mathematical struc-
ture of the real number line, so too, the empirical structure of belief states
can be represented by the logico-linguistic structure of propositions.
Assuming, then, that (TA) is true, nominalistic scientific realism
becomes very plausible. For if all the platonistic talk appearing in empirical
science is a mere descriptive aid-or an aid to our descriptions and our under-
standing of the physical world-then there doesn't seem to be any good
reason to believe that it's true, because fictions can aid our descriptions and
our understanding as easily as truths can. (As a case in point, consider that
the novel Animal Farm could very easily serve as a descriptive aid in an his-
torical account of the years surrounding the Russian Revolution. We can say
something roughly true about Stalin by uttering the sentence, 'Stalin was
like the pig Napoleon,' even though this sentence is, strictly speaking, false.
In other words, we can coherently believe the historical content of this sen-
tence without believing its Animal-Farm content.) And the same is true in
empirical science: we could use abstract-object talk to accurately characterize
the operation or state of some part of the physical world even if there were no
such things as abstract objects.20 Therefore, the fact that we do use abstract-
object talk in this way does not provide any reason whatsoever to think that
this talk is true. And this is why I think that nominalistic scientific realism
is a plausible view, why I think it is just as plausible as ordinary scientific
realism; because none of the reasons for endorsing the nominalistic content of
empirical science provides any good reason to endorse its platonistic content.
To put the point as starkly as possible, the reason nominalistic scientific
realism is a sensible philosophy of science is that the nominalistic content of
empirical science is all empirical science is really "trying to say" about the
world; its platonistic content is something it "says incidentally" in its effort
to say what it really "wants to say". Or to lapse completely into metaphor,
the nominalistic content of empirical science is its picture of the physical
world, whereas its platonistic content is the canvas (or part of the canvas) on
which this picture is painted; thus, in order to endorse empirical science's pic-
ture of the physical world, we needn't endorse its platonistic content.
I end this section by considering an objection one might raise against the
position I've been developing here, an objection that might be put in
20 One might worry that the presence of fictional material in our empirical theories could
"infect"the nominalistic content of those theories. But so long as the fictitious entities are
not taken to be causally efficacious, this cannot happen.
AYTITUDES WITHOUT PROPOSITIONS 819
something like the following way. "If what you've argued is correct, then
whenever I have a belief, i.e., whenever I hold up my end of some belief
ascription, it will be because my brain is in some particular state. This
means that part of my brain-presumably a neural sequence, or state, or
event, or some such thing-will have a physicalistic content-determining
property, i.e., a physicalistic property which makes it the case that I have the
particular belief in question, rather than some other belief. But to claim that
some neural state in my brain has a content-determining property is just to
claim that it has content, or meaning. Thus, fictionalists are committed to
the thesis that neural states are meaningful, and so they are going to have to
account for this. But you now face a dilemma. If, as platonists would have it,
we need to countenance propositions to account for meaningfulness-the idea
being that propositions just are sentence meanings-then believing that p
will involve being related in a certain way to a proposition after all, and so
fictionalism will be false. If, on the other hand, we can account for the mean-
ingfulness of neural states without committing to propositions, then the
motivation for fictionalism will have evaporated, for we will have stumbled
onto an anti-platonist view that enables us to avoid making the wild
fictionalist claim that belief ascriptions like (3) are false. In particular, we
will be able to say that content-bearing neural states are the referents of 'that'-
clauses and the objects of belief. In other words, we will be able to say that
mental representations, or sentence tokens of the neural language of thought
(Mentalese), are the referents of 'that'-'clauses."21
First of all, fictionalists are not committed to Mentalese sentences. They
are committed to the claim that whenever a person x holds up her end of an 'x
believes that p' sentence, there is some purely nominalistic fact about her
brain (and perhaps her environment) that makes this the case. But they
needn't maintain that in such cases there is a Mentalese token stored in x's
brain which (in platonistic terms) means that p. (And even if they allow that
in some cases, there is such a token stored in x's head, they might very well
deny that there is in all cases.22) But in the next section, I will argue that
even if fictionalists grant the Mentalese-token view of the physical facts of
belief, they can still avoid both horns of the above dilemma. I will begin
with the first horn, explaining how fictionalists can respond to the platonist
worry that we need to countenance propositions to account for the meaning-
fulness of Mentalese tokens. Then, in response to the second horn, I will
point out that even if fictionalists can and should appeal to Mentalese tokens
21 For a defense of this view, see JerryFodor, The Language of Thought(Cambridge, Mas-
sachusetts:HarvardUniversity Press, 1975).
22 Actually, this some-but-not-all view is probably the strongest version of the Mentalese
view that could be correct. For we surely don't want to explain the fact that Madonna
believes that 434 > 17 by claiming that there is a Mentalese token in her head that means
that 434 > 17.
820 MARK BALAGUER
in explaining how humanshave beliefs, there are still good reasons to doubt
that Mentalese tokens are the referents of 'that'-clauses and the objects of
belief.
4. A FICTIONALIST ACCOUNT OF MEANINGFULNESS

I begin by explaining how fictionalists can respond to the Frege-Katz
argument that we need propositions to account for the meaningfulness of
public-languagesentences, and then I extend the storyto Mentalesesentences.
We can see how fictionalists can respond to the Frege-Katz argument by
noticing that it is a special case of the Frege-Bealer-Schifferargument.To
claim that we need propositionsto accountfor the meaningfulnessof public-
language sentences is just to claim that we need them to accountfor the truth
of certain 'that'-clause-containingmeaningascriptions,e.g.,
(5) The sentence token that Floyd just uttered meant that 0. J. is a
bachelor.
Now, it might seem that we could make the problem for fictionalists
worse by appealingnot just to sentences like (5), but also to sentences like
(6) The sentence type 'O. J. is a bachelor' means that 0. J. is a bach-

elor.
For it seems that in order to account for the truth of (6), we need to coun-
tenance two kinds of abstractobjects, viz., propositions and sentence types.
Platonistsmight try to buttressthis point by arguingthat any adequatemean-
ing theory for English would have to be about sentence types, rather than
tokens, because-in orderto be adequate-it would have to entail, for every
English sentence a, a theoremof the form 'a means that p'. But fictionalists
have a response to this. Let MTE be a meaning theory for English of the
above sort, i.e., one that makes claims about all English sentences and,
hence, must be taken as being about sentence types. Theories like MTE and
sentences like (6) are purely platonistic, ratherthan mixed; they are, thus,
more akin to the sentences and theories of mathematics than to those of
physics and belief psychology, and so fictionalists can take the same line
with them that they take with mathematics.In other words, fictionalists can
grant that sentences like (6) and theories like MTE are "about"propositions
and types but maintainthat since there are no such things as propositions or
types, such sentences and theories are completely fictional. Now, in orderto
block this move, platonists would need to construct an argument for the
claim that sentences like (6) and theories like MTE are true. It seems to me
that the only promising strategyhere would be to constructan indispensabil-
ity argumentanalogous to the one they use to argue that mathematical sen-
tences and theories are true-i.e., an argumentthat shows that we need to
suppose that sentences like (6) and theories like MTE are true in order to
account for some class of facts about the physical world. But the only facts
thatplatonistsmight reasonablyappealto here are facts aboutactual physical
tokens-e.g., the fact (or alleged fact) expressed by (5). Thus, even if we
begin by talking about (6) and MTE, it seems clear that what fictionalists
really have to do is respondto the worrythat we need propositionsto account
for the truthof sentences like (5).
Let me turn, then, to this worry. My suggestion is that fictionalists can
say the same thing about the meaning ascription(5) that they say about the
temperatureascription(4) and the belief ascription(3). Since the proposition
that 0. J. is a bachelor is causally inert, it follows that the mixed fact
expressedby (5) superveneson two (morebasic) facts which aren't mixed and
which are entirely independentof one another-viz., a purely physical fact
about Floyd's token (and the surroundingcommunity)and a purely platonis-
tic fact about the proposition that 0. J. is a bachelor. But from this it fol-
lows that (5) has a nominalistic content that capturesits complete picture of
Floyd's token-a nominalistic content that says that Floyd's token holds up
its end of the "(5) bargain"-and thatit is perfectlyreasonableto believe this
nominalistic content while maintainingthat the platonistic content of (5) is
fictional.
Now, in order to make this view plausible, I need to say what kind of
purely physical facts underlie sentences like (5). To do this, we need to dis-
tinguish two different facts that (5) might be taken to express, i.e., that (5)
conflates. If Floyd was speaking literally when he uttered his token, then
these two facts are distinguishedby
(5e) The token that Floyd just uttered meant in English that 0. J. is a
bachelor
and
(5u) The token that Floyd just uttered meant to us that 0. J. is a bach-
elor.
The purely physical facts underlyingthese two sentences are different. The
physical fact behind (5e) is similar in spirit to the physical fact behind the
assertionthat Floyd's utterancewas a syntacticallywell-formed English sen-
tence token. That is, it is a fact aboutthe shape of Floyd's token. In particu-
lar, it is the fact that Floyd's token is shaped in such a way that makes it
hold up its end of the "Floyd's-token-is-a-token-of-a-type-that-means-in-
English-that-0. J.-is-a-bachelorbargain".And the purely physical (or neuro-
logical, or psychological, or whatever)fact behind (5u) is the fact thateach of
us associates Floyd's token with a "mentalrepresentation"of the proposition
822 MARK BALAGUER
that 0. J. is a bachelor, where a "mentalrepresentation"is a neural token of
some sort.23
The reason I put scare quotes around'mentalrepresentation'is to empha-
size that I am using this term as I have been using 'about' throughoutthe
paper: on my usage, a neural token can be said to "represent"a proposition
even if, in point of actualfact, there are no such things as propositions.And,
again, the reason it is acceptable(and, indeed, helpful) to speak in this way is
that propositions (if there exist such things) are causally inert. For given
this, it follows that I could construct "mental representations"and assign
them to tokens even if there were no propositions for my so-called
"representations"to be representationsof. This, of course, is just to reiterate
that the fact that I associate Floyd's token with a certain "mentalrepresenta-
tion" is a purely physical (or neurological, or psychological, or whatever)
fact. And it is this which, in turn, suggests that (5u) has a nominalistic con-
tent that capturesits complete pictureof us and Floyd's token-a nominalis-
tic content that says that (a) all of us, and (b) Floyd's token, hold up our two
ends of the three-way"(5u)bargain".
Now, of course, this view of (5u) presupposesthat we have some anti-pla-
tonistic story to tell about the meaningfulness of Mentalese tokens. For to
say that a neural token "represents"the proposition that p is just to say that
it has this proposition as its content-i.e., that it means that p. (After all, in
orderto count as "representing"the propositionthatp, ratherthan some other
proposition, a neural state is going to have to have a content-determining
property.)Thus, we are back where we were at the end of section 3, needing a
response to the platonist claim that we need propositions to account for the
truthof sentences like
(5m) The particular neural token n means in Mentalese that 0. J. is a

bachelor.
23 Two points. First, on some sense of '...means to us...', there are (a) sentences which
mean things to us but which some of us don't understand(and so it would not be the case
that each of us associates the sentence with an appropriatemental representation) and
(b) sentences which mean things to us but which have never been associated with any
"mental representation",because they have never been consciously considered. (E.g.,
'439.2 > 14.6' might be such a sentence-or, at least, it might have been before I just
considered it.) But this notion of '...means to us...' is equivalent to '...means in our lan-
guage...', and so it can be handled along the lines of (5e). The notion of '...means to
us...' that I have in mind in connection with (5u) applies only when a token is actually
cognized and understoodto mean something by a particularperson or group of persons.
Second, note that a token can mean to us something other than what it means in
English. For what "mentalrepresentation"a person x assigns to a token t is a function of
(a) what t means in English and (b) context, i.e., factors such as whether x thinks t was
uttered metaphorically,whether x heard t correctly, etc.
But it ought to be very clear by now what fictionalists can say about (5m).
Since the proposition that 0. J. is a bachelor is not causally related to n, it
follows that the mixed fact expressed by (5m) superveneson two non-mixed
facts that are entirely independentof one another,viz., a purely physical fact
about n and a purely platonistic fact about the proposition that 0. J. is a
bachelor. And from this it follows that (5m) has a nominalistic content that
captures its complete picture of n-a nominalistic content that says that n
holds up its end of the "(5m) bargain"-and that it is perfectly coherent to
believe this nominalistic content while maintaining that the platonistic
content of (5m) is purely fictional.
Now, one might wonder what kind of purely physical facts underlie sen-
tences like (5m), i.e., what the physicalisticcontent-determiningpropertiesof
neural tokens are like. Unfortunately,given our currentignorance of neuro-
physiology, it is impossible to say anything very informative here. But this
is not a problem for fictionalists qua fictionalists, because (a) platonists also
need to solve this problem, and (b) fictionalists can give any solution to it
that platonists can give. The reason platonists are on all fours with
fictionalists here-why they too owe an account of the physicalistic content-
determiningpropertiesof brain states, or neural tokens-is simply that they
too are committed to the claim that brain states do have such properties,
because they too are committedto the claim that we form "mentalrepresenta-
tions" of propositions. Now, I suppose that one might doubt that platonists
are committedhere to the existence of physicalistic content-determiningprop-
erties, but it is easy to see that they are. For since propositions (if there are
such things) are causally inert, our brains are going to have to form their
"representations" on their own, so to speak, i.e., without receiving any
"help"from any abstractobjects.
(It is importantto keep in mind here that by 'physicalistic', I mean anti-
platonistic, ratherthan anti-mentalistic. Thus, platonists cannot help their
cause here by claiming that they can avoid having to believe in physicalistic
content-determiningpropertiesby adoptinga mind-braindualism and claim-
ing that while our Minds can construct representationsof propositions, our
brains cannot-or at least do not. For, ignoring the fact that this sort of dual-
ism is implausible, fictionalists can make the same move. That is, if platon-
ists could somehow motivate dualism, fictionalists could accept the view
with them, because dualism is consistent with anti-platonism.One can main-
tain that it is our Minds that construct"mentalrepresentations"but still deny
that there are any such things as propositions.)
In any event, if what I've arguedis correct,then fictionalistscan avoid the
first horn of the dilemma presented at the end of section 3: it may well be
that platonism is the only view that can account for the (alleged) truth of
meaning ascriptions like (5) and (6), but there is no good argumentagainst
the view which denies that such sentences are true, which takes (6) (and
824 MARK BALAGUER
MTE) to be completely fictional and (5) (and (5e), (5u), and (5m)) to have
true nominalistic contents (that captureeverythingimportantthat they entail
about the physical world, i.e., about tokens and speakers) but fictional pla-
tonistic contents.
But what aboutthe second horn of the dilemma?One might think that the
above argumentssimply show that platonism is, in fact, not the only view
that can accountfor meaningfulness,or more generally,for the truthof 'that'-
sentences. One might think the above arguments show that anti-platonists
can account for the truth of 'that'-sentences by taking "mental representa-
tions" to be the referentsof 'that'-clauses,i.e., the meanings of sentences, the
objects of belief, etc.
The fact of the matter,though, is that the above argumentsshow nothing
of the sort. All I've argued is that "mentalrepresentations"can be used to
explain certain sorts of facts, e.g., the fact that Floyd holds up his end of the
"(3) bargain"and the fact that we hold up our end of the "(5u) bargain".But
there is a big difference between the claim that "mentalrepresentations"can
be put to this minimal use and the claim that they are the referentsof 'that'-
clauses. Indeed, while the former claim seems very plausible, there are
numerouswell-known argumentsfor thinkingthat the latter claim is false-
namely, the argumentsthathave been offered by platonistslike Bealer, Schif-
fer, and Katz. I alluded to these argumentsin section 2 in discussing the sec-
ond premise of the Frege-Bealer-Schifferargument,i.e., the premise that the
only plausible suggestion about what 'that'-clausesmight refer to is the pla-
tonistic suggestion that they refer to propositions. Unfortunately,I haven't
the space to go into these argumentshere,24but it is worth noting that none
of them provides any reason whatsoever to doubt that "mental representa-
tions" can be used in the way thatfictionalists want to use them, i.e., to
explain how it is that people can hold up their ends of 'that'-sentences like
(3) and (5u).
This last point capturesthe real beauty of the fictionaliststance: it enables
anti-platoniststo ignore the argumentsof people like Bealer, Schiffer, and
Katz. We can admit that 'that'-sentencesare best interpretedas being "about"
propositions without admittingthat there are propositions. More generally,
we can admitthat our best theories are best interpretedplatonistically-either
as purely platonistic theories or as mixed theories-and we can reap all the
benefits of interpretingthem in this way without committingto the existence
of abstractobjects. Seen in this light, anti-platonistsshould not lament hav-
ing no option but to endorse fictionalism, because it seems to be the most
attractiveversion of anti-platonismwe've got, because it enables us to have
24 Well, let me give one very quick argument.Look at (5m). Its 'that'-clause couldn't refer
to a neural token, because if it did, the neural token in question would have itself as its
meaning.
AYTITUDES WITHOUT PROPOSITIONS 825
our cake and eat it too. On the one hand, fictionalists don't have to "watch
what they eat" like other anti-platonistsdo; they can make full use of our
seemingly platonistic theories without worrying about whetherevery entail-
ment of these theories can be given anti-platonistictruthconditions. But on
the otherhand,they don't have to pay any price for their "recklesseating",as
platonists do with their ontological obesity.25
25 Here is an interesting aside. If everything I've said here is correct, then the standard
relationalview of the metaphysics of belief is wrong. To have a belief is not to be related
in a certain way to the referent of a 'that'-clause, because we can have beliefs even if
'that'-clauses fail to refer-i.e., even if there are no such things as objects of belief. For
this reason, it seems to me that having a belief is a one-place property of believers
(although it may be that what beliefs a person has is determined not just by her neural
state but also by external factors). Now, it might seem odd that I would accept a one-
place metaphysics of belief, because I have already said that I endorse a two-place
semantics of 'believes'. But there is really nothing odd about this; semantics and meta-
physics come apart all the time, because we very often use "metaphysically irrelevant
entities" as descriptive aids to express facts. For instance, the best semantic analysis of
ordinary (Celsius) temperatureascriptions holds that such sentences are true if and only
if the physical system in question is Celsius-relatedto the real numberin question. But this
does not mean that to have a temperaturejust is to be related in a certain way to a real
number.Likewise, while 'believes' is a two-place predicate,having a belief is not a two-
place relation. And it seems to me that this is true even if there do exist propositions,
because propositions are not metaphysically relevant to our belief states. But of course,
much more would need to be said to defend this wedding of the standard two-place
semantics of 'believes' to a one-place metaphysics of belief.
826 MARK BALAGUER

Balaguer 1998

Uploaded by

Document Informationclick to expand document information

Document Informationclick to expand document information

Copyright:

Available Formats

Balaguer 1998

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Balaguer 1998

Uploaded by

Copyright:

Available Formats

International Phenomenological Society

Philosophy and Phenomenological Research

Attitudes without Propositions

You may need to log in to JSTOR to access the linked references.

California State University, Los Angeles

This paper develops a novel version of anti-platonism,called semanticfictionalism.The

806 MARK BALAGUER

ATTITUDES WITHOUT PROPOSITIONS 807

Jenny believes thatIdgie is a dog

B(j, the referentof 'thatIdgie is a dog')

(2) That 0. J. is a bachelorentails that 0. J. is unmarried

are false (or not true).

(3) Floyd believes that Bill Clinton is president

to explain human behavior.

810 MARK BALAGUER

3. A FICTIONALIST ACCOUNT OF BELIEF

ATrTrUDES WITHOUT PROPOSITIONS 811

(NC) Empiricalscience has a purely nominalisticcontent that capturesits

(COH) It is coherent and sensible to maintainthat the nominalistic con-

812 MARK BALAGUER

(4) The physical system S is forty degrees Celsius.

ATlTUDES WITHOUT PROPOSITIONS 813

814 MARK BALAGUER

ATTITUDES WITHOUT PROPOSITIONS 815

816 MARK BALAGUER

(TA) Empirical theories refer to abstract objects (most notably mathemati-

ATTITUDES WITHOUT PROPOSITIONS 817

818 MARK BALAGUER

AYTITUDES WITHOUT PROPOSITIONS 819

820 MARK BALAGUER

4. A FICTIONALIST ACCOUNT OF MEANINGFULNESS

(6) The sentence type 'O. J. is a bachelor' means that 0. J. is a bach-

ATTITUDES WITHOUT PROPOSITIONS 821

822 MARK BALAGUER

(5m) The particular neural token n means in Mentalese that 0. J. is a

ATTITUDES WITHOUT PROPOSITIONS 823

824 MARK BALAGUER

AYTITUDES WITHOUT PROPOSITIONS 825

826 MARK BALAGUER

You might also like