Maxdeutsch.files.wordpress.com

The One and Only Argument for Radical Millianism The Southern Journal of Philosophy (2006) Vol. XLIV The One and Only Argument for
Radical Millianism
Max Deutsch
East Carolina University
Abstract
Radical Millianism agrees with less radical varieties in claiming thatordinary proper names lack “descriptive senses” and that thesemantic content of such a name is just its referent but differs fromless radical varieties of Millianism in claiming that any pair ofsentences differing only in the exchange of coreferential names cannotdiffer in truth-value. This is what makes Radical Millianism radical.
The view is surprisingly popular these days, and it is popular despitethe fact that, until very recently, there was not a single argument for it.
Theodore Sider and David Braun (2006) have tried to provide themissing argument, but, I argue, their attempt fails. I conclude that we(still) have no reason to be Radical Millians.
Not every philosopher of language who is convinced by thepowerful antidescriptivist arguments of Saul Kripke’s Namingand Necessity is willing to adopt what I shall call RadicalMillianism, a view on the semantics of names articulated anddefended by Nathan Salmon (1986), Scott Soames (1987, 2002),and more recently by David Braun (1998) and Michael Thau(2002). Radical Millianism says that ordinary proper nameslack “descriptive senses” and that the semantic content of sucha name is just its referent, but also, and most radically, it saysthat any pair of sentences differing only in the exchange ofcoreferential names cannot differ in truth-value. It is a conse-quence of Radical Millianism that the following statements (1)and (2), for example, (must) share a truth-value: Max Deutsch is assistant professor of philosophy at East Carolina University in Greenville, North Carolina. He earned his doctoratefrom Rutgers University in October of 2001. He has recently publishedin Synthese and is the coauthor of a texbook on philosophical logicpublished by Continuum. His current interests include the semanticsof names and the attitude reports, intentionalist theories of conscious-ness, and the semantics-pragmatics interface. (1) Lois Lane believes that Superman can fly.
(2) Lois Lane believes that Clark Kent can fly.
There are, it would seem, many good reasons not to adopt the Radical Millianism of Salmon and company. There is, ofcourse, the almost universal judgment that (1) and (2) do differin truth-value. But there is also the fact that there are other,less radical ways of being an antidescriptivist. What I shall callNon-Radical Millianism says that names are not descriptiveand that their semantic contents are just their referents, butdoes not follow Radical Millianism when it comes to substitu-tivity: An exchange of coreferential names might, according tothe Non-Radical Millian, make for a difference in truth-value.
Various versions of Non-Radical Millianism have been developed.
Mark Crimmins and John Perry (1989) have argued that,although names are nondescriptive devices of “direct reference,”belief reports like (1) and (2) implicitly refer to different con-textually salient modes of presentation and that this differenceexplains the reports’ differing truth-values. Similarly, MarkRichard’s (1990) theory of names and belief reports allows, forreasons quite different from those offered by the “hidden-indexical” theory of Perry and Crimmins, that (1) and (2) maydiffer in truth-value, while concurring with Radical Millians onthe nondescriptive nature of names. Why be Radical when thereare plausible Non-Radical alternatives? As far as I can tell, there is exactly one argument for going Radical. This argument, the Argument for Radical Millianism(ARM, for short), is due to Theodore Sider (1995) and is rela-tively new. Very recently, it has been re-presented in a papercoauthored by Sider and David Braun (2006). According to Siderand Braun, ARM shows that no correct theory of names andbelief reports can allow the truth of both “S believes that a is F”and “S does not believe that b is F,” where “a” and “b” are coref-erential proper names. For example, ARM allegedly shows that(1), from above, and (3), below, cannot be true together.1 (3) Lois Lane does not believe that Clark Kent can fly.
One clear target of ARM is descriptivism. By allowing thatdifferent descriptive senses may attach to “Superman” and“Clark Kent,” descriptivist accounts of the semantics of namesallow that (1) and (3) can both be true. More interestingly,however, ARM also targets Non-Radical Millianism. Just asNon-Radical Millianism allows that (1) and (2) may differ intruth-value, it also allows that (1) and (3) may be jointly true.
But it opens the door to these possibilities without postulatinga difference in descriptive sense between “Superman” and“Clark Kent.” Non-Radical Millianism is thus a satisfying, The One and Only Argument for Radical Millianism middle-of-the-road sort of view. If we adopt it, we get to agreewith the antidescriptivism of Naming and Necessity, but wealso get to trust our judgments about the truth-values of beliefreports. But, if ARM is sound, Non-Radical Millianism holdsout false hopes. For if ARM is sound, there is no such compro-mise position, our judgments about the truth-values of beliefreports are simply wrong, and we must all convert to RadicalMillianism.2 Although I am not, myself, at all attracted to Radical Millianism, it is, from a certain perspective, something of arelief that there is finally an argument for the theory. Despiteits problems (the clash with pretheoretic judgment, the exis-tence of more attractive seeming options), Radical Millianismhas received a fair amount of good press and continues toattract converts, or at least those willing to expend some energydefending the view. For example, quite a lot of ink has beenspilled recently detailing the ways in which the judgments thatconflict with Radical Millianism might be “explained away,”perhaps by appealing to “pragmatic factors.” But why beattracted to the theory of Radical Millianism in the first place?Surprisingly little has been said about this. In fact, I think it isfair to say that, until now, not a single positive reason forendorsing Radical Millianism has ever been offered.3 ARM, ifsound, would remedy this situation.
1. The Argument
Sider has presented ARM twice, once in a critique of MarkRichard’s theory of names and belief reports (Sider 1995) and,more recently, in the 2006 paper coauthored with David Braun,a paper in which the authors critically discuss Scott Soames’sBeyond Rigidity (2002). The more recent presentation is easierto follow and less enmeshed in intricacies peculiar to Richard’stheory, so it is this version that I will discuss here. But theearlier presentation is fuller in the sense that the argument’spresuppositions are made explicit and certain objections aredealt with, so I will draw from the earlier presentation whenhelpful as well.
The more recent (Sider and Braun 2006) version of ARM specifically targets the possibility that (1) and (3) from aboveare both true. (Sider (1995) calls all theories that allow thispossibility, including Non-Radical Millian theories, “Fregean,”noting that the use is nonstandard, since some such theoriesare antidescriptivist. My own use of “Fregean” below should beinterpreted in this Siderean, nonstandard way as well.) Here ishow ARM appears in the 2006 paper.4 The Argument for Radical Millianism (ARM)
Superman = the superhero & Lois Lane believes thatSuperman can fly Therefore, ᭚x (x = the superhero & Lois Lane believes that xcan fly) Clark Kent = the milquetoast reporter & Lois Lane does notbelieve that Clark Kent can fly Therefore, ᭚x (x = the milquetoast reporter & Lois Lane doesnot believe that x can fly) Therefore, ᭚x (x = the superhero & Lois Lane believes that xcan fly & Lois Lane does not believe that x can fly) (from b, d,and e) Sider and Braun claim that the argument’s final line, f, is
contradictory (2006, 678). In support of this, they point out that
“the more nearly English rendering” (678) of f is given by f*:
Therefore, there is someone who is the superhero, and is suchthat Lois Lane both believes that he can fly and does notbelieve that he can fly.
Sider and Braun say that f* “cannot be interpreted as expressing
anything other than a contradiction” (2006, 679). Let’s suppose
—though we will return to the issue below—that they are right
about this: f/f* is a contradiction. How does its derivation from
a–e show that (1) and (2) cannot be true together?
(1) is the second conjunct of a and (3) is the second conjunct
of c. It is common ground that (in the context of the stories)
Superman = the superhero and that Clark Kent = the milque-
toast reporter, and so common ground that the first conjuncts of
a and c are true. Hence, if the contradiction is ultimately trace-
able to a and c, it is their latter conjuncts, that is (1) and (3),
that are to blame. Is the contradiction ultimately traceable to a
and c? It certainly looks that way: e is also common ground,
and the move from b, d, and e to f involves the substitution of
one description for another, codenoting description in a non-
intensional context—a valid move, surely. That leaves the moves
from a to b and from c to d, but these seem perfectly legiti-
mate. In each case, the inference is from an instance of a
certain existential generalization to that generalization itself.
How could such inferences fail to be valid? As Sider puts it,
“[o]ne wonders why existentially quantified sentences like [b
and d] make sense at all if they never follow from particular
instances; after all, one teaches the use of the existential quanti-
The One and Only Argument for Radical Millianism fier by pointing out that any instance whatsoever is logicallysufficient for the truth of an existential sentence” (Sider 1995,511). And thus it seems that ARM is just what it purports to be:a reductio of the claim that (1) and (3) can be true together.
2. There Must Be Something
Wrong with ARM
It is tempting to reply to ARM by insisting that, despite initialappearances, there must be some misstep in it, since it isplain—plainer, even, than the validity of the straightforward-seeming inferential principles on which ARM relies—that (1)and (3) can be, and are, both true. But this is question beggingat worst, and at best it shows merely that there is a clash ofintuitions regarding names and belief reports: On the one hand,we intuit that (1) and (3) can be true together, but, on the other,we intuit (or so Sider and Braun would say) that ARM is valid,and thus that (1) and (3) cannot be true together. Nevertheless,ARM does, I think, “prove too much,” even if pointing to the factthat it “proves” that (1) and (3) are inconsistent is not thefairest way to show that it does.
A more effective way to make the “proves too much” point is to note that nothing in ARM seems to limit its scope to beliefreports containing proper names (and other alleged devices of“direct reference”) in their “that-clauses.” In particular, nothingin the argument seems to bar us from using codenoting definitedescriptions to play the same role as that played by coreferen-tial names in the original version. Hence, if ARM has a validform, we ought to be able to use it to prove that the reports (4)and (5) below, which contain codenoting definite descriptionsinstead of coreferential proper names in their that-clauses,cannot be true together: (4) Lois Lane believes that Lex Luthor’s nemesis can fly.
(5) Lois Lane does not believe that Perry White’s lackey can fly.
But if ARM proves that (4) and (5) cannot be jointly true, then itreally does prove too much. That (4) and (5) can both be true ismore than a mere intuition, and every actual theory of descrip-tions and belief reports (and every remotely plausible possibletheory of such things) allows for such a possibility. Certainly,any theory that claimed that descriptions function like Milliannames, contributing only their denotations to the meanings ofbelief reports containing them (thus ruling out the possibilitythat (4) and (5) are both true), would be rejected by any sensibletheorist. Different descriptions, even if codenoting, typically andobviously differ in their semantic contribution to their con-taining sentences, and hence can, if interchanged, make for a difference in the truth-values of those sentences. But, as the i–
vi variant on ARM below shows, ARM appears to “go through”
even if we replace the coreferential “Superman” and “Clark
Kent” with the codenoting “Lex Luthor’s nemesis” and “Perry
White’s lackey.”5
ARM: Descriptions Version
Lex Luthor’s nemesis = the superhero & Lois Lane believesthat Lex Luthor’s nemesis can fly Therefore, ᭚x (x = the superhero & Lois Lane believes that xcan fly) iii. Perry White’s lackey = the milquetoast reporter & Lois Lane
does not believe that Perry White’s lackey can fly Therefore, ᭚x (x = the milquetoast reporter & Lois Lane doesnot believe that x can fly) vi. Therefore, ᭚x (x = the superhero & Lois Lane believes that x
can fly & Lois Lane does not believe that x can fly) (from ii, iv,and v) For reasons similar those described in the discussion of the
original ARM, in this Descriptions Version, the apparently
contradictory vi (which is just f, renamed) appears to follow
from the seemingly unobjectionable i–v. And for the same
reasons that, in ARM, the contradiction appeared to stem
ultimately from lines a and c, in the variant, the source seems
to be a and c’s analogues, namely i and iii. But it is common
ground that (in the stories) Lex Luthor’s nemesis = the super-
hero and that Perry White’s lackey = the milquetoast reporter;
hence, it seems that the second conjuncts of i and iii, that is (4)
and (5), respectively, are to blame. However, we know already
that this cannot be right; (4) and (5) clearly can both be true.
What the Descriptions Version shows, then, is that something,
somewhere, has gone awry—both in it and in ARM.
3. But What?
Part 1: Contradictions and Intuitions
ARM must be flawed. But how is it flawed? One problem is
that, despite what Sider and Braun say, it is not obvious that
the ARM’s final line, f, expresses a contradiction.6 Sider and
Braun, remember, say that f’s contradictoriness is plain when
we consider its “more nearly English rendering,” f*:
The One and Only Argument for Radical Millianism Therefore, there is someone who is the superhero, and is suchthat Lois Lane both believes that he can fly and does notbelieve that he can fly.
But whether f/f* is contradictory depends on the semantic
interpretation of the occurrences of the variable, “x,” in f, and
the pronoun, “he,” in f*. Perhaps the proper interpretation of
the variable/pronoun involves a process whereby variables, or
their natural language equivalents (e.g., pronouns), have
descriptive meanings attached to them.7 Perhaps, for example,
the only sense in which f follows from a–e is a sense according
to which f means roughly what f** does.
f** Therefore, there is someone who is the superhero and is such
that Lois Lane both believes that he, the superhero, can flyand does not believe that he, the milquetoast reporter, can fly.
But f** sounds true, not contradictory.8 So if f** is all that
follows from a–e, then a–e do not imply that (1) and (3) are not
possibly true together.
What could rule out a descriptive interpretation of the variable/pronoun in f/f*? Presumably only an antidescriptivist
theory of the semantic content of variables and pronouns would
do the trick. But the trouble with adopting such a theory in the
context of presenting an allegedly theory-independent argument
against all Fregean theories of names and belief reports—and
thus, in particular, against Fregean descriptivism—ought to be
obvious: The descriptivist’s descriptivism about names will
presumably extend to variables and pronouns (just as the
Millian’s Millianism typically extends from names in this way).9
In other words, if someone, Sider or Braun perhaps, were to
insist that the variable/pronoun in f/f* does not have a descrip-
tive interpretation, then what looked to be an interesting argu-
ment against all Fregean theories of names and belief reports,
and looked not to make any explicitly anti-Fregean assumptions
of its own, would turn out, upon closer inspection, to depend on
a kind of anti-Fregeanism about variables and pronouns. And
that, of course, would make ARM far less interesting than it
initially appears.
However, Sider and Braun say very little by way of explain- ing why we ought to view f/f* as a contradiction. They do say
that it “cannot be interpreted as expressing anything other
than a contradiction” (2006, 679), but this is not a clear endorse-
ment of anti-Fregeanism about variables and pronouns, and
given that such an endorsement would considerably weaken
their argument, perhaps we ought to consider other interpreta-
tions of their claim.
Sider and Braun’s case for f/f*’s being a contradiction may
be simply that it would strike an ordinary speaker (or hearer) as such. If this is what Sider and Braun mean when they say
that f* cannot be interpreted as anything other than a contra-
diction, then, although it is not really a matter that can be
settled from the armchair, my guess is that they are mistaken.
My guess is that ordinary speakers would even sometimes fail
to judge (6), below, as a contradiction.
(6) Lois believes that Superman can fly and Lois does not believe One reason they might fail to so judge is that, even given aMillian construal of names, (6) is ambiguous, and on only onereading is it contradictory. (6)’s latter conjunct is a “negativebelief report,” that is, a sentence of the form “A does not believethat p.” In general, such reports suffer from scope ambiguity:The negation in them can take large scope—“It is not the casethat: A believes that p”—or small—“A believes that not-p.”Small scope readings of negative belief reports do not contradictthe corresponding positive belief report; the conjunction “Abelieves that p and A believes that not-p” attributes contra-dictory beliefs to A but is not itself contradictory. Hence, evengranting Millianism for names, (6) has a reading according towhich it says that Lois both believes that Superman can fly andthat Superman cannot fly. This reading is not inconsistent; itsays merely that Lois has contradictory beliefs. The fact that (6)has this reading suggests that ordinary speakers will notalways judge (6) to be contradictory, or even false.
It ought to be fairly obvious how the issue of scope ambi- guity bears on the question of how ordinary speakers will judge
f*. f*—“Therefore, there is someone who is the superhero and is
such that Lois both believes that he can fly and does believe
that he can fly”—is a (quantified) conjunction, one conjunct of
which is (something close to) a negative belief report, namely,
“Lois does not believe that he can fly.” Hence, even if the
pronoun in f* is interpreted in a Millian way, f* is semantically
ambiguous, and on only one of its disambiguations is it contra-
dictory. If the negation in its latter conjunct takes small scope
over the belief operator (in that same conjunct), then, assuming
Millianism for pronouns, f* says (roughly) that there is a
superhero about whom Lois has contradictory beliefs: she
believes that he can fly and that he can’t. Saying that is consis-
tent. So it is possible, likely even, that many ordinary speakers
would not take f* to express a contradiction. They might hear
only its noncontradictory, negation-gets-small-scope reading.
Suppose we first train those we ask about f*’s contradictoriness
to recognize scope ambiguities, and we insist that they take the
negation in its latter conjunct as having scope over the belief
operator (in that conjunct). Would they then find f* contra-
The One and Only Argument for Radical Millianism dictory? My guess is that they would not. My guess is that even
when given analogous instructions for interpreting (6), many
ordinary speakers would balk at the claim that (6) is a contra-
diction. Many ordinary speakers know that Superman keeps his
secret identity hidden from Lois, and, for such speakers,
describing one result of Superman’s deception by uttering (6),
even when instructed on how to interpret the negation in its
latter conjunct, would probably seem perfectly consistent—
accurate even. The same, I suspect, is true of f*. Even if taken
to be the claim that there is someone who is the superhero and is
such that Lois both believes that he can fly and fails to believe
that he can fly, f* will strike many as perfectly consistent.
However, to the extent that the intuitions of the ordinary speaker are to be trusted about such issues, what matters is not
an intuition about f*’s contradictoriness (or lack of it) consi-
dered, as it were, in isolation. Considered in isolation, there are,
as I have been stressing, many ways to interpret f*. What we
ought to be curious about is an intuition about f* contra-
dictoriness (or lack of it) given that it follows from a–e. That is,
if, as Sider and Braun suggest, the ordinary speaker will intuit
that ARM is valid (and this, I think, is a very big “if ”—see the
next section) will they also intuit that its conclusion, f*, is a
contradiction? Again, this is an empirical matter, but I’m fairly
confident that the answer will be “no.” The reason, of course, is
that most ordinary speakers will intuit that a and c—
“Superman = the superhero & Lois Lane believes that Superman
can fly” and “Clark Kent = the milquetoast reporter & Lois
Lane does not believe that Clark Kent can fly”—are both true.
But a and c are ARM’s only genuine assumptions, the lines
from which all the other lines follow. So if the ordinary speaker
intuits that a and c are true and intuits that the argument is
valid, then the ordinary speaker will, if he or she genuinely
understands the notion of validity, conclude that f* must be not
only noncontradictory, but true!10
4. But What?
Part 2: The Failure of
Existential Generalization
Even if we were to grant that f/f* is, and is intuitively, a
contradiction, there would still be a serious flaw in ARM. The
argument is not valid: b does not follow from a, and, for the
same reason, d does not follow from c.
Superman = the superhero & Lois Lane believes thatSuperman can fly Therefore, ᭚x (x = the superhero & Lois Lane believes that xcan fly) Clark Kent = the milquetoast reporter & Lois Lane does notbelieve that Clark Kent can fly Therefore, ᭚x (x = the milquetoast reporter & Lois Lane doesnot believe that x can fly) b and d are existential generalizations of a and c, respectively,
but existential generalization, a rule of inference that, if
unrestrictedly valid, would license the moves from a to b and
from c to d, is known to fail on terms within the scope of an
intensional operator like “believes-that.”11 Indeed, part of what
it is to be an intensional operator is for existential generaliza-
tion to fail on terms within the operator’s scope. Of course, one
potential reason for the failure of existential generalization in
intensional contexts is the possible “emptiness” of the instantial
term. “Someone is such that Ann believes that he is a spy”
follows from “Ann believes that the present king of France is a
spy” only if “the present king of France” is nonempty.12 But this
reason does not apply to the inferences from a to b and c to d;
both “Superman” and “Clark Kent” (we are pretending) refer.13
Still, even adding the condition that the relevant terms be
nonempty, existential generalization is not valid in intensional
contexts. The issue here is similar to the old issue of when, if
ever, a “de re” belief report can be inferred from the corre-
sponding “de dicto” one. Even given that there exists a unique
shortest spy, the de re (8) does not follow from the de dicto (7).
(7) Ann believes that the shortest spy is a spy.
(8) The shortest spy is such that Ann believes that he is a spy.
As David Kaplan would put it, (8) expresses information thatwould be of interest to the FBI: It implies that there is aparticular someone whom Ann believes to be a spy. (7), on theother hand, carries no such implication. Indeed, (7) might betrue simply because Ann arrived at the belief it ascribes to hervia the general beliefs that spies exist and that no two of themshare a precise height. Ann’s reasoning in this way is not,however, sufficient grounds for the truth of (8). The same pointswould seem to apply to (9), which is the existentialgeneralization of (7), and the question of whether it followsfrom (7).
(9) Someone is such that Ann believes that he is a spy.
Like (8), (9) implies that there is a particular someone whomAnn believes to be a spy. As is the case with (8), (9)’s truth isnot a trivial consequence of Ann’s reflections on general truthsabout spies and their heights. Hence, even granted that thereexists a unique shortest spy, (9) cannot be inferred from (7). The The One and Only Argument for Radical Millianism more general lesson, then, is that existential generalization
fails on terms within the scope of an intensional operator, even
if that term refers. A consequence of the general lesson would
seem to be that it is true neither that b follows from a nor that
d follows from c, and, hence, that ARM is not valid.14
Nonetheless, I can imagine at least two reasons why someone sympathetic to ARM might find the discussion in the previous
paragraph not quite relevant to the question of the argument’s
success. First, one might complain that all that the previous
discussion shows is that existential generalizations of belief
reports do not follow from those reports themselves. It does not
show that de dicto belief reports in combination with certain
auxiliary assumptions do not entail their generalizations. And,
the complaint might continue, there is no reason to suppose that
these auxiliary assumptions, whatever they may be, could not be
true in the circumstances in which a and c are true (and thus
that, given a and c, and given these auxiliary assumptions, b
and d follow). Second, one might complain that the examples of
the previous paragraph involve a definite description and that
what we say about existential generalization and de dicto-to-de
re inferences might be different when the examples involve
proper names. Perhaps inferring the existential generalization,
“Someone is such that Lois believes that he can fly,” or the de re,
“Superman is such that Lois believes that he can fly,” from the de
dicto, “Lois believes that Superman can fly,” is unproblematic in
a way in which analogous inferences from a de dicto belief
report with a definite description in its “that-clause” are not.
This second complaint comes dangerously close to begging the crucial question. Take the issue of de dicto-to-de re entail-ment (a.k.a. the issue of “exportation” in belief contexts). It hasbeen pointed out by some of Radical Millianism’s defenders thatif Radical Millianism is true, then there is no de dicto/de redistinction with respect to belief reports containing propernames in their that-clauses.15 The de dicto report, “Lois believesthat Superman can fly,” and the de re report, “Superman is suchthat Lois believes that he can fly,” ascribe a belief in one andthe same “object-involving” “singular” proposition to Lois; hence,the reports’ truth-conditions and content are exactly the same.
So the answer to the question of whether the former reportentails the latter is, trivially, “yes.” As stressed, the reasoninghere depends on assuming Radical Millianism. Analogousreasoning for the conclusion that there is a trivial deductiveconnection between name-containing de dicto belief reports andtheir existential generalizations would presumably require thesame assumption. But that assumption is obviously notavailable to the proponent of ARM. ARM is meant to be anargument for that assumption.16 The first complaint—that there may be certain auxiliary assumptions that, when true, would allow an inference from a de dicto belief report to its existential generalization—would
amount to the second, if it were accompanied by the claim that
the relevant auxiliary assumption is just that the instantial
term be a proper name. This version of the first complaint falls
prey to the charge of question begging described in the previous
paragraph. Is there some more neutral set of auxiliary assump-
tions that would, in combination with the truth of a de dicto
report, imply that report’s existential generalization? David
Kaplan has famously argued that de re belief reports are
inferable from the corresponding de dicto reports, given certain
further assumptions having to do with the subject of the
report’s epistemic connection to the object of his or her belief. If
there is some such set of further assumptions, their being true
would clearly also allow inferences to existential generalizations
of de dicto belief reports from those reports themselves.
Perhaps, if Lois is sufficiently “en rapport” with Superman, her
believing de dicto that Superman can fly implies that there is
someone such that Lois believes that he can fly. However, doubts
have been raised about whether Kaplan has succeeded in
specifying the correct set of assumptions, and some philos-
ophers have gone as far as claiming that the project of seeking
such assumptions is ill conceived. There are no such further
assumptions, these skeptical philosophers say, and to think
there are is to misunderstand the nature of de re belief. De re
belief, they claim, is fundamental, and not even partially
analyzable in terms of de dicto belief. If this sort of skepticism
about the Kaplanian project is justified, then that presumably
closes the book on ARM: b does not follow from a (nor d from c)
and there are no further assumptions, short of b (or d) itself,
that would bridge the gap.
However, even supposing that exportation and existential generalization are allowable in certain special circumstances isnot going to rescue ARM; ARM requires something more specificthan this. First, it requires the validity of both “positive” and“negative” existential generalization; that is, it must be possibleto move both from a positive belief report of the form “Abelieves that a is F,” and from a negative belief report of theform “A does not believe that b is F,” to their respective existen-tial generalizations—that is, to reports of the form “There issomething such that A believes that it is F” and “There issomething such that A fails to believe that it is F.” But, second,ARM also requires that positive and negative existentialgeneralization be legitimate even when the instantial terms arecoreferential.17 The trouble with these more specific require-ments is that no one can meet them; the Fregean can’t, but thenneither can the Radical Millian.
The reason these more specific requirements cannot be met is best approached via the related issue of the invalidity ofpositive and negative exportation. If one holds that in certain The One and Only Argument for Radical Millianism circumstances one can infer the “positive” de re form, “a is suchthat A believes that it is F from the “positive” de dicto one, “Abelieves that a is F,” then one cannot further hold that, in thosesame circumstances, and where “a” and “b” corefer, one can alsoinfer the “negative” de re form, “b is such that A fails to believethat it is F” from the “negative” de dicto form “A fails to believethat b is F.” Allowing such a thing would lead to contradiction:“a is such that A believes that it is F and a is such that A failsto believe that it is F.”18 One might reply that this is, in effect,Sider and Braun’s point: contradictions result from allowing thetruth of both “positive” and “negative” de dicto belief reportsinvolving coreferential proper names. But the trouble is withexportation and existential generalization, and it is not peculiarto names but afflicts belief reports containing coreferentialterms quite generally. If Jones believes that the mayor is acrook, but fails to believe that his neighbor is a crook, and themayor = Jones’s neighbor, then, if exportation is valid in boththe positive and negative cases, contradictions will result. Thisshows that at least one of the relevant inferences is not valid,not that Jones cannot believe that the mayor is a crook whilesimultaneously failing to believe that his neighbor is a crook.19In sum, there are general reasons, having nothing in particularto do with one’s favored semantics for proper names, fordisallowing the possibility on which ARM depends, that is, fordisallowing cases of simultaneous positive and negativeexistential generalization where the instantial terms arecoreferential.20 5. Conclusion
Radical Millianism needs a positive argument. In my view, thereis currently far too much enthusiasm for a semantic theory thatwe have never been given reason to believe.21 Following thepublication of Naming and Necessity, the situation, as I seematters, was this: Descriptivism of the Frege/Russell varietyhad been very convincingly laid to rest. However, and as Kripkehimself has stressed, the question of what, positively, to sayabout the semantics of names, and especially what to say abouttheir behavior in attitude contexts, remained an open question.
As far as I can tell, this is still, to this day, a completely openquestion. There is simply nothing in Kripke’s work that suggestsor implies the truth of Radical Millianism. Surprisingly, there isalso nothing in the work of present-day Radical Millians thatcounts as a reason for believing the theory they endorse. Part ofthe problem is that present-day Radical Millians have busiedthemselves with a defensive project: “Given the truth of ourtheory, what can we say to defend it from the evidence thatappears to conflict with it?” This sort of attitude is perhaps anatural one, especially given the nature of the evidence against Radical Millianism. After all, the theory conflicts not just withesoteric philosophical principles but with robust, easily elicitedintuitions had by nearly everyone. Even so, such defenses won’t(or ought not) convince anyone who is not already a RadicalMillian. Radical Millians need now to go on the offensive. We allknow that false theories can be given ingenious defenses. Weneed to be shown that Radical Millianism is a theory worthgiving such a defense.
ARM looked promising in this regard, and Sider and Braun are to be commended for at least making an attempt to arguepositively for Radical Millianism. But I believe I have hereexposed ARM as a failure. The question, then, is whether thereis something that can be put in its place. This is the challengeto Radical Millians. Until it is met, we should be suspicious ofthe value of defenses of Radical Millianism. Berkelian idealismcan be defended too.22 More precisely, what ARM allegedly shows is that (1) and (3) cannot be true together, when (3) is read as the negation of (2). Ifsound, ARM also shows that (1) and (2) cannot differ in truth-value,again taking (3) as (2)’s negation. There is a scope ambiguity in (3);this is why the claims about what ARM allegedly shows must bequalified. The issue of (3)’s scope ambiguity will reappear later in themain text. It forms the basis of one of my criticisms of ARM.
Sider and Braun think that ARM not only shows that Radical Millianism is true, but also that it can be extended to show that the“Millian Descriptivism” elaborated in Scott Soames’s Beyond Rigiditycannot be correct. That is, they believe that ARM can be used notonly to show that we have to be Radical Millians, but also that wehave to be Radical Millians of a particular variety. We can’t, accordingto them, be Radical Millians who maintain that descriptive proposi-tions are pragmatically imparted by utterances of name-containingbelief reports. (This, very roughly, is Soames’s Beyond Rigidityposition.) To me, the real interest of ARM is its potential to show thatno version of Fregeanism could be correct. Whether it can be used towhittle down the range of possible versions of Radical Millianism isvery much a side issue. However, if my criticisms of ARM below arecorrect, then Sider and Braun are wrong in thinking that it can beextended, in the way they imagine, to criticize Soames’s MillianDescriptivism. (Note 14, below, briefly explains why.) Having said that,I should add that the ARM-based criticism of Soames’s MillianDescriptivism is only one of many raised by Sider and Braun (2006).
I suspect that many philosophers take Kripke’s arguments in Naming and Necessity, and perhaps especially in “A Puzzle AboutBelief,” to be pro-(Radical-)Millian. This is a mistake. Those argumentsare antidescriptivist and decidedly not pro-Millian. As Scott Soamesreminds us in Beyond Rigidity, in Naming and Necessity, Kripke toldus what the semantic content of a name cannot be, not what it is. Infact, Kripke says things in Naming and Necessity that, if true, entailthe falsity of Millianism, or at least of Radical Millianism. (He says The One and Only Argument for Radical Millianism that “Hesperus is Hesperus” is a priori but that “Hesperus isPhosphorus” is a posteriori, e.g.) As for “A Puzzle About Belief,” themain thrust of the argument in that paper is that a certain powerfulseeming anti-Millian argument is not as powerful as it seems. This toofalls far short of an argument for, or even an endorsement of,Millianism. Some Millians seem to think that pointing to cases inwhich a pair of coreferential names seem, intuitively, to be inter-changeable in certain attitude contexts somehow counts as arguing fortheir view. This is like arguing for the validity of affirming theconsequent by pointing out that there are some cases in which whenone reasons from “If p, then q,” and “q,” to “p,” one is not led from truthto falsehood. Radical Millianism says that, in every case and in anycontext, an exchange of coreferential names does not lead to adifference in truth-value. That there are some cases and contexts inwhich an exchange of coreferential names does not alter truth-value isirrelevant. Despite this, Millians like Bryan Frances (2002) claim to“prove” that Radical Millianism is true by describing cases in whichcertain well-chosen examples of coreferential names do seem freelyinterchangeable in certain belief contexts. But even a Fregeandescriptivist can allow that some coreferential names can be freelyinterchanged, namely those with the same descriptive sense. Francesis not the only culprit. David Braun (1998) claims that “some of ourintuitions about belief reports actually support Russellianism[“Russellianism” is Braun’s label for Radical Millianism.]” (Braun1998, 559). His defense of this claim is that “we often take a beliefreport to be true as long as it ‘gets the reference right’” (Braun 1998,559). And he goes on to give an example in which swapping anindexical for a (coreferential) name in the “that-clause” of a beliefreport seems, intuitively, not to affect the truth of the report. Howcould this possibly count as support for Radical Millianism? Of course,sometimes, coreferential terms can be interchanged within the contentclauses of belief reports salva veritate. But this would be true even ifevery name, indexical, or other allegedly directly referential term weresynonymous with some definite description or other. I class MarkRichard’s famous phone-booth argument (Richard 1983) as an exampleof the same sort of faulty reasoning. Richard argues that there arecases in which it is intuitive that certain belief ascriptions involvingdifferent but coreferential terms in their that-clauses (In Richard’soriginal case, the terms are indexical.) have the same truth–value andcontent. Again, the existence of such cases is irrelevant to the truth ofMillianism; there would be such cases even if Millianism were false.
Sider and Braun call ARM “A4” (2006, 677–78). My lettering of the premises and conclusion of ARM matches theirs.
Some readers may be less of a nerd than I am: Lex Luthor is the arch-villain of the Superman stories. Perry White is the editor of TheDaily Planet, the newspaper for which Clark Kent, a.k.a. Superman,works as a reporter.
f and vi are identical, so what I say about f will of course apply
to vi as well.
Sider and Braun describe an analogous process—“descriptive enrichment”—whereby variables/pronouns have descriptive contentassociated with them, but only at the pragmatic level. What I’mproposing is descriptive enrichment, but at the semantic level.
Or suppose that the proper interpretation of f* is: “Therefore
there is someone who is the superhero and is such that Lois Lane both
believes that he—when thinking of him as ‘the superhero’—can fly and
does not believe that he—when thinking of him as ‘the milquetoast
reporter’—can fly.” This sounds even more clearly true than f**.
It is somewhat surprising, from a historical perspective, that Sider and Braun imply that the variable/pronoun in f/f* is plainly
Millian, and that this can be exploited in cooking up a powerful anti-
descriptivist, pro-Millian argument. Nearly twenty-five years ago,
Alonzo Church (1982) suggested that the fact that statements like f/f*
follow from statements like (1) and (2) (modulo some uncontroversial
auxiliary assumptions) is precisely a reason to deny that the variable/
pronoun in f/f* is Millian. Church would say that this counts as a
reason to take the variable/pronoun in f/f* as possessing a Fregean
sense. Church’s papers on quantification and the propositional
attitudes are missing from the bibliographies of (Sider 1995) and
(Sider and Braun 2006).
10 Sider and Braun make something of a fuss over the difference between what they call “particular intuitions,” i.e., intuitions about thetruth-values of sentences, and what they call “logical intuitions,” i.e.,intuitions about the validity of arguments. “Logical intuitions,” theywrite, “are not merely subservient to intuitions about particularsentences. They are independent, and indeed have the potential toclash with particular intuitions” (2006, 674). One of their maincontentions is that while particular intuitions favor Fregean (in Sider’ssense) theories, some logical intuitions appear to favor RadicalMillianism. In particular, the intuition that ARM is valid is alleged tofavor Radical Millianism. However, the point about the “independence”of logical and particular intuitions is only partly true. It is true thatone can make a judgment about the validity of an argument withouthaving an opinion about the truth-values of the argument’s premisesand conclusion. Logical intuitions and particular intuitions areindependent in this sense. But if one has the particular intuition that“S” is true and the particular intuition that “T” is false, then the twointuitions combined just is, in some sense, the “logical intuition” thatthe argument from “S” to “T” is not valid. In a way, then, particularand logical intuitions are not independent. The point bears on Siderand Braun’s claim about our intuitive reactions to ARM: If we dointuit the truth of the argument’s premises and the falsity of itsconclusion, as they insist we do, then haven’t we thereby intuited thatthe argument is invalid? If so, it is misleading for Sider and Braun todescribe the situation to be such that it is only particular intuitionsthat tell against Radical Millianism. There is also a logical intuition,based on (subservient to?) the particular intuitions about its premisesand conclusion, that ARM is invalid. Sider and Braun seem to thinkthat we will also have a separate intuition to the effect that theargument is valid. This strikes me as unsupported speculation, buteven if it is right, there is a question about how much weight thisseparate intuition will be given. A reason to suspect that it will begiven little weight is that it is probably true in general that particularintuitions are firmer than logical intuitions, especially when thearguments to which the logical intuitions apply are relatively complex(as is the case with ARM).
11 Unrestricted existential generalization says: From any sentence, “… a …,” where “a” is a singular term, infer “᭚x (… x …).” The One and Only Argument for Radical Millianism 12 This seems, at any rate, to be a plausible constraint on existen- tial generalization. The matter is somewhat controversial, however.
13 In any case, the first conjuncts of a and c imply the existence of
14 Sider and Braun (2006) claim that ARM constitutes an objec-tion not just to Fregean theories but also to a certain version of Radical
Millianism, like Soames’s “Millian Descriptivism” (see note 2, above).
However, ARM can be extended in this way only if we accept that (A)
ARM is intuitively valid, and (B) f/f* is a contradiction. But I’ve
argued here that we should reject both (A) and (B). Hence, I don’t
think that ARM poses a threat to Millian Descriptivism.
15 When this is pointed out, it is usually in the spirit of identifying some of the interesting consequences of Radical Millianism. Althoughthis is more properly a topic for another paper, my view is that RadicalMillianism’s inability to “see” de dicto/de re ambiguities counts asfurther evidence against the theory. I think there are clear cases inwhich the truth conditions of “S believes that a is F” and “a is suchthat S believes that it is F” differ, even when “a” is a proper name. Seenote 16, below, for a presentation of one such case.
16 Here again it may be that Sider and Braun are simply relying on a supposed intuition to the effect that the existential generalizations
of de dicto belief reports containing names (b and d) do follow from
those reports themselves (a and c). I do not have such an intuition,
and I suspect that, to the extent that the ordinary speaker does, it is
also had with respect to de dicto belief reports containing (denoting)
definite descriptions, and hence should be taken with a grain of salt.
Anyone who has the intuition that name-containing de dicto belief
reports straightforwardly imply the corresponding de re reports can
come to see the error of his or her ways by consider a case involving a
name whose reference is “fixed by description.” Suppose Superman
arrives at a murder scene and catches a glimpse of the horribly
mutilated body of Smith. Somehow, it is clear that there is only one
perpetrator. Superman introduces the name “Alfred” for the murderer,
whoever it may be, of Smith. He then forms the belief that Alfred is
insane. It ought to be obvious, however, that nothing said so far suffices
for its being true that Alfred is such that Superman believes that he is
insane. Given the obvious similarities between exportation and
existential generalization, these sorts of cases ought to make us
discount intuitions to the effect that exportation involving (nonempty)
names is valid.
17 How does ARM depend on these further requirements? The move from a to b involves positive existential generalization, while the move
from c to d involves negative existential generalization. But
“Superman” and “Clark Kent” are coreferential.
18 Assuming, that is, that the relevant variables/pronouns are 19 Sider (1995, 512) claims that intuition tells us that both positive and negative exportation are unrestrictedly valid. Any thought to thecontrary results, Sider claims, from overexposure to early work onexportation and de re belief (e.g., Quine 1956 and Kaplan 1968). Butanyone who judges that Jones can believe that the mayor is a crookwhile failing to believe that his neighbor is a crook, even when Jones’sneighbor = the mayor, can be made to see that maintaining thisjudgment requires denying that both positive and negative exportation are valid. And they can be made to see this without exposing them tothe details of philosophical theories on the topic.
20 Sider (1995) imagines a Fregean responding to ARM by claiming, perhaps on the basis of commonalities between de re and existentiallygeneralized versions of de dicto belief reports, that there is an“asymmetry” between positive and negative existential generalization.
One can, Sider imagines his Fregean critic saying, validly inferexistential generalizations of positive de dicto reports from thosereports themselves, but one cannot validly infer existential generali-zations of negative de dicto reports from the reports themselves (Sider1995, 511). This is not the view I mean to be defending here. There isno asymmetry; neither positive nor negative existential generalizationis valid. Though it is doubtful, it may be that, under certain conditionsperhaps having to do with the relevant believer’s epistemic connectionto the object of his or her belief, one may infer a positive generaliza-tion from the corresponding positive de dicto report. By the sametoken, it may be legitimate, again in certain special circumstances, toinfer a negative existential generalization from the correspondingnegative de dicto report. What has been argued here is that whatcannot be allowed, under any circumstances, is both positive andnegative existential generalization where the instantial terms arecoreferential. Positive existential generalization, considered alone, isno better or worse than negative existential generalization consideredalone. It is their combination that causes trouble.
21 One of the negative effects of this overenthusiasm is the philo- sophical time and energy spent on problems that would not arise inthe absence of a commitment to Millianism. Explaining the persistenceof objects is a real problem only if you assume Berkelian idealism.
Similarly, a particularly virulent form of the “problem of empty names”arises only by assuming Millianism: “Since empty names have nocontent, given that names must refer in order to have content, howcan their use in truth-evaluable assertions be explained?” One wonderswhether we should fret much about this version of the empty namesproblem in the absence of some positive reason for endorsing theMillianism it presupposes. At least Berkeley argued for his idealism.
22 I’d like to thank John Collins, Harry Deutsch, Nick Georgalis, Wong Pak Hang, Pei Kong Ngai, Joe Lau, and Dan Robbins for helpfuldiscussions of the material in this paper. Special thanks are due toMichael Veber who, besides discussing the paper’s content with me,gave me some very useful advice, which I took, concerning its style.
References
Braun, D. 1998. Understanding belief reports. Philosophical Review Church, A. 1982. A remark concerning Quine’s paradox about modality.
Crimmins, M., and J. Perry. 1989. The prince and the phone booth: Reporting puzzling beliefs. Journal of Philosophy 86:685–711.
Frances, B. 2002. A test for theories of belief ascription. Analysis Kaplan, D. 1968. Quantifying In. Synthese 19:178–214.
Kripke, S. 1979. A puzzle about belief. In Salmon and Soames 1988, The One and Only Argument for Radical Millianism 102–48. Originally published in A. Margalit, ed., Meaning and use(Dordrecht: D. Reidel, 1979), 239–75.
Kripke, S. 1972/1980. Naming and necessity. Oxford: Basil Blackwell.
Quine, W. V. O. 1956. Quantifiers and propositional attitudes. Journal Richard, M. 1990. Propositional attitudes: An essay on thoughts and how we ascribe them. New York: Cambridge University Press.
Richard, M. 1983. Direct reference and ascriptions of belief. In Salmon and Soames 1988, 169–96. Originally published in the Journal ofPhilosophical Logic 12:425–52.
Salmon, N. 1986. Frege’s puzzle. Cambridge: MIT Press.
Salmon, N., and S. Soames, eds. 1988. Propositions and attitudes. New Sider, T. 1995. Three problems for Richard’s theory of belief ascription.
Canadian Journal of Philosophy 25:487–514.
Sider, T., and D. Braun. 2006. Kripke’s revenge. Philosophical Studies Soames S. 1987. Direct reference, propositional attitudes, and semantic content. Philosophical Topics 15:47–87.
Soames, S. 2002. Beyond rigidity: The unfinished semantic agenda of Naming and Necessity. New York: Oxford University Press.
Thau, M. 2002. Consciousness and cognition. New York: Oxford

Source: http://maxdeutsch.files.wordpress.com/2009/07/theoneandonlyargumentforradicalmillianism.pdf