GRAHAM OPPY

SEMANTI CS FOR PROPOSI TI ONAL ATTI TUDE

ASCRI PTI ONS*
(Received 4 April, 1991)
Nathan Salmon (1986a) (1986b) (1989) and Scott Soames (1987)
(1988) defend a neo-Russellian account of the semantics of proposi-
tional attitude ascriptions in which senses and modes of presentation
play no part. They also defend an under-developed pragmatic theory in
which senses and modes of presentation do play an important role.
Graeme Forbes (1987a) (1987b) defends a neo-Fregean account of
the semantics of propositional attitude ascriptions in which senses and
modes of presentation play a crucial part. He contends that the view
which is defended by Salmon and Soames is not really an alternative to
the view which he defends. (Hence, I suppose, his view is not an
alternative to their view, either.)
There are two issues which I wish to take up. First, in sections I--V,
I explore the question whether there are really any important differ-
ences between the view defended by Salmon and Soames, and the view
defended by Forbes. The conclusion which I defend is that there is very
little which distinguishes between them.
Second, in sections VI--X, I take up the somewhat neglected ques-
tion of how theories of propositional attitude ascriptions in the Frege-
Russell tradition ought to construe the notions of sense and mode of
presentation.
In his review of Salmon (1986a), Forbes (1987b) claims that it is
possible to see the theory which Salmon advances in that book as a
notational variant of the neo-Fregean theory of propositional attitudes
and propositional attitude ascriptions which Forbes (1987a) defends.
Against this, Jose Branquinho (1990) claims that: "[While] there must
be s ome sense in which Salmon's account could be seen as invoking the
Philosophical Studies 67:1--18, 1992.
9 1992 KluwerAcadernic Publishers. Printed in the Netherlands.
GRAHAM OPPY
same kind of conceptual apparatus as the neo-Fregean theory . . .
Forbes "notational variation claim cannot be accepted." (p. 22) I think
that Branquinho is correct to argue that the actual claim which Forbes
defends in his review is defective. However, I also think that the
intuition which motivated Forbes' claim is perfectly sound. Sections
I I - - V of mY paper are a defence of this position.
lI
If we allow ourselves a little terminological freedom, we may represent
one central claim which Salmon (1986a) defends as follows:
(S) ~ believes that p-7 is true iff (3x) (A grasps that Russeltian
proposition that p by means of the Salmonian guise x and
BEL (A, the Russellian proposition that p, the Salmonian
guise x))
where BEL is a ternary relation which holds bet ween possessors of
propositional attitudes, Russellian propositions, and ways of grasping
Russellian propositions ( = Salmonian guises).
Similarly, we may represent the Fregean postulate for which Forbes
(1987b) claims equivalence as follows:
(F) ~ is believed by A to obt a i ~ is true iff (3x) (A thinks of the
state of affairs that p by means of the Fregean proposition x
and B (A, the Fregean proposition x) and the state of affairs
that p is the content of the Fregean proposition x)
where B is a binary relation which holds bet ween possessors of propo-
sitional attitudes and Fregean propositions.
If we: (i) identify Russellian propositions with states of affairs; (ii)
identify Fregean propositions with Salmonian guises; (iii) identify the
relation of grasping-a-Russellian-pr~176176176176
guise with the relation of thinking-of-a-state-of-affairs-by-means-of-a-
Fregean-proposition; and (iv) identify the relation BEL (A, p, x) with
the product of the relations B (A, x) and C (p, x) - - where C (p, x) iff
the state of affairs that p is the content of the Fregean proposition x - -
then the right hand sides of (S) and (F) are also identical. In other
words, given these apparently unproblematic identifications, Salmon's
SEMANTICS FOR ATTITUDE ASCRIPTIONS 3
theory attributes the same truth-value to a sentence rA believes that p7
which Forbes theory attributes to the sentence ~ is believed by A to
obtain ~ (and vice versa).
However, as Branqhinho in effect points out, this does not show that
the two theories are notational variants. The crucial difficulty is that,
even after the proposed series of identifications, the left-hand sides of
(S) and (F) differ. In a familiar -- though perhaps objectionable --
vocabulary, one might say this: Salmon's account entails the same
distribution of truth-values to de dicto propositional attitude ascriptions
which Forbes' account gives to the corresponding de re propositional
attitude ascriptions. Less problematically, we can certainly say that the
theories of Salmon and Forbes disagree about the truth-values which
are assigned to sentences of the form ~ believes that p~. Hence, there
is no way that the theories can be construed to be notational variants
when the only thing which is taken into account is the distribution of
truth-values to sentences which is made by these theories.
IlI
Despite the fact that Forbes' actual argument is incorrect, the intuition
which motivates his argument is sound. As his translation scheme
suggests, the two theories invoke exactly the same theoretical entities.
Hence, the differences between the two theories must be due to the
different deployment of these entities in the theories. Thus, a charge of
(something like) notational variation could still be sustained if it could
be shown that these entities actually play the same sort of role in each
theory, but in different locations in those theories.
The crucial difference between the two accounts of sentences of the
form ~A believes that 1 ~v lies in the fact that Salmon's guises play a role
in pragmatics which Forbes' Fregean propositions play in semantics.
On Forbes' theory, Fregean propositions play a role in the recursive
assignment of truth-values to sentences. On Salmon's theory, Salmonian
guises do not play this role. However, this is not to say that Salmon's
guises do not play a structurally similar role in the pragmatic com-
ponent of Salmon's theory. It seems to me that they must.
In order to account for the fact that his theory seemingly yields an
assignment of truth-values to sentences which is wildly at variance with
4 GRAHAM OPPY
pre-t heoret i cal intuition, Sal mon claims t hat t hose pre-t heoret i cal intui-
tions are actually intuitions about the correct assertability of proposi -
tional at t i t ude ascriptions. In particular, he claims t hat proposi t i onal
at t i t ude ascriptions carry (convent i onal or generalised) Gri cean implica-
tures about t he Sal moni an guises under which Russellian proposi t i ons
are ent ert ai ned.
Sal mon tells us al most not hi ng else about how these Gri cean impli-
catures get at t ached to sentences. However, it is as obvious t hat t here
must be a composi t i onal t heor y involving Sal moni an guises which issues
in an assignment of assertability-values to sent ence-cont ext pairs as it is
t hat t here must be a composi t i onal meani ng t heor y for languages which
issues in an assi gnment of truth-values to sent ence-cont ext pairs. (How
else coul d we account for the fact t hat speakers can recognise the
assertability-values of novel sent ences? How else coul d speakers have
the ability to pr oduce and under st and a pot ent i al l y infinite range of
sentences with at t ached assertability-values?) Moreover, it is equally
clear t hat this t heor y will have exactly the same st ruct ure as the neo-
Fregean t heor y of Fregean propositions. That is, it is clear t hat what
Salmon' s t heor y does is to shift some of the st ruct ure which is f ound in
the Fregean t heor y from semantics to pragmatics.
(This concl usi on is likely to appear unpal at abl e to direct reference
theorists. Af t er all, t here are oft en supposed to be i nsuperabl e difficul-
ties which face the const ruct i on of a Fregean semantic theory. How-
ever, it seems to me t hat if this were right, it could not be a wel come
concl usi on for direct reference theorists, since it woul d show t hat t hey
are not be able to cont ruct the pragmat i c theories which t hey need in
or der to account for the way we actually speak.)
Neo-Russel l i ani sm and neo-Fregeani sm are structural variants: the
same recursive structures involving the same underl yi ng entities appear
in each theory. However - - at least pri ma f aci e - - the two theories are
not not at i onal variants; for t hey disagree on t he assi gnment of truth-
values to sentences, and t hat certainly seems to be a substantive differ-
ence bet ween them.
IV
The consi derat i ons advanced in II may seem t o const i t ut e an argument
SEMANTICS FOR ATTITUDE ASCRIPTIONS 5
for neo-Fregeani sm. Af t er all, given t hat t he onl y di fference bet ween
the two t heori es lies in t he l ocat i on of the recursive st ruct ure involving
Fregean proposi t i ons, surely we shoul d opt for the t heor y whi ch gives a
di st ri but i on of t rut h-val ues (and assertabillty-values) which is in accord-
ance wi t h pre-t heoret i cal intuition.
Well, I' m not so sure. The quest i on is whet her we really have any
reason to hol d t hat t he relevant pre-t heoret i cal intuitions are a reflec-
t i on of the truth-values whi ch are possessed by proposi t i onal at t i t ude
ascriptions, or whet her t hey are merel y a refl ect i on of t he assertability
of t hose sentences. The two t heori es seem to be on a par i n every
respect - - i.e. t here is no simplicity argument whi ch favours one rat her
t han the ot her - - so we need to deci de whet her t he relevant intuitions
are themselves di rect evi dence which favours one of the alternatives, or
whet her t hey woul d onl y const i t ut e such evi dence if t here were some
furt her ar gument which gave us good reason for thinking t hat t hose
intuitions do support the view whi ch t hey seem to support. (I don' t see
t hat it is any use saying: look, Loi s Lane really doesn' t know that Clark
Kent is Superman. For each of t he t heori es predicts t hat this is what we
shall say.)
Ther e are clear cases in whi ch phenomena shoul d be assigned to
pragmatics. (Consi der, for example, t he phenomena which Gri ce dis-
cussed in his original papers on implicature.) Ther e are also clear cases
in which phenomena shoul d be assigned to semantics. (Simple subject-
predi cat e sentences are surely of this sort, since in their case t here is no
r oom for t he sort of di vergence bet ween t rut h and assertability which
di rect reference theorists claim to find in the case of proposi t i onal
at t i t ude ascriptions.) However, in proposi t i onal at t i t ude ascriptions, t he
fact t hat t here are two distinct el ement s (object and sense) which must
be br ought i nt o pl ay means t hat we need some argument to tell us
where is the appropri at e l ocat i on of these factors in our t heory.
Perhaps t here is an argument which shows t hat the neo-Russellians
are definitely wrong. However, even if t here were such an ar gument it
woul d prove little, since t here is so little di fference bet ween t he two
views. (Note, in particular, t hat t he views agree on all the predi ct i ons
which t hey make about which proposi t i onal at t i t ude ascriptions speak-
ers will and will not endorse. ) Neo-Fregeans and neo-Russellians woul d
do bet t er to wor r y about their common probl ems - - e.g. empt y singular
6 GRAHAM OPPY
terms, the nature of modes of presentation -- than to worry about their
alleged differences.
v
Forbes is well aware of the fact that his theory and Salmon's theory do
not agree on the assignment of truth-values to sentences of the form
believes that p~. Why then did he give the faulty argument which I
discussed in section I above? The answer to this question is, I think,
contained in the opening paragraph of his review. There, he says that
propositions are "things which are the meanings of sentences and the
objects of propositional attitudes" and then adds that: "The important
questions are (a) according to the best philosophy of language, are the
meanings of sentences Russellian propositions or Fregean ones and (b)
does the most plausible philosophy of mind take the objects of the
attitudes to be Russellian propositions or Fregean ones?" If there is to
be just one entity which is both the object of propositional attitudes and
the meaning of the content clauses of propositional attitude ascriptions,
then the fact that there is a sense in which neo-Fregeans and neo-
Russellians take the same objects to be the objects of propositional
attitudes is good evidence that those theories are notational variants.
(Of course, neo-Russellians will say that the objects of propositional
attitudes are Russellian propositions. But when it comes to the explana-
tion of behaviour, they agree that one has to appeal to Fregean proposi-
tions -- i.e. to modes of presentation of Russellian propositions. So, in
the most important sense, they agree that the objects of propositional
attitudes are Fregean propositions.)
But, of course, in the sense in which neo-Fregeans and neo-Russel-
lians take the same objects to be the objects of propositional attitudes,
it is simply incorrect to attribute to the neo-Russellian the view that
there is just one entity which is both the object of propositional
attitudes and the meaning of the content clauses of propositional
attitude ascriptions. (In general, I think that one should be very careful
in talking about "t he meaning" of sentences or types of sentences.
There are various different things which make good candidates for the
meanings of sentences, depending upon the theoretical purposes at
hand.) The claim that neo-Russellianism and neo-Fregeanism are nora-
SEMANTICS FOR ATTITUDE ASCRIPTIONS 7
tional variants relies on Forbes' neo- Fr egean assumpt i on t hat the
objects of proposi t i onal attitudes which are i nvoked in the expl anat i on
of behavi our are also the semantic correlates of the relevant proposi -
tional at t i t ude ascriptions. But this assumpt i on is not shared by direct
reference theorists (nor by ot her neo- Fr egean theorists).
vI
Suppose t hat we are agreed t hat we want a proposi t i onal analysis of
proposi t i onal at t i t ude ascriptions of the sort which is offered by neo-
Russellians and neo-Fregeans. (This is a big assumption; however, I
don' t want t o go i nt o that here.) Then t here are some very i mpor t ant
questions about modes of present at i on which need to be addressed.
For definiteness, I shall adopt a neo- Fr egean not at i on (though, as I
explained above, I do not suppose t hat anyt hi ng much hangs on this).
Since t he formal aspects of this semantic t heor y are well- known, I shall
onl y provi de a very quick outline:
The first thing which we do is to descri be the proposi t i ons which are
assigned to sentences relative to a cont ext and an assi gnment of individ-
uals to variables:
( i ) t he proposi t i on expressed by an at omi c formul a r - Gb l . . , b ~
relative to a cont ext c and an assi gnment f is (G*, (b~ . . . . ,
bn)), where G* is t he pr oper t y expressed by F-G~ in the
cont ext c, and b~ is t he individual which is the semantic
cont ent of r-b~ relative to c and f.
(ii)
the pr oposi t i on expressed by '~-b believes that A-~, relative to
c and f, is ((b, Prop A, Mode A), B), where B is the bel i ef
relation, b is the cont ent of rb~ relative to c and f, Prop A is
the pr oposi t i on expressed by r-A~ relative to c and f, and
Mode A is a mode of present at i on of the proposi t i on whi ch
is expressed by r-A~ relative to c and f.
Next, we give a st andar d account of the intensions of expressions
relative to contexts, and use this to provi de a recursive charact eri sat i on
of t rut h relative to ci rcumst ances of evaluation:
8 G R A H A M O P P Y
O )
(ii)
a proposi t i on (G*, (bl, 9 9 9 b.)) is t rue relative to a circum-
stance of eval uat i on E iff the ext ensi on of G* in E contains
( b 1 . . . . , b n ) -
a proposi t i on ((b, Prop A, Mode A), B) is t rue relative to a
ci rcumst ance of eval uat i on E iff (b, Prop A, Mode A) is a
member of the ext ensi on of B in E - - i.e. iff b believes Prop
A under Mode A in E.
Of course, we need to say a lot mor e about the nat ur e of modes of
presentation. It seems reasonabl e to suppose t hat a mode of present a-
tion of a proposi t i on will be a tuple of the modes of present at i on of the
constituents of t hat proposition. Thus, we shall have:
( i ) a mode of present at i on of a proposi t i on {G*, {bl, 9 9 9 bn) is
a tuple ([G*], [bl], . . . , [bn]), where [G*] is a mode of
present at i on of t he pr oper t y G* and [ b i ] is a mode of
present at i on of the individual b i.
a mode of present at i on of a proposi t i on {{b, Prop A, Mode
a ) , B) is a tuple {[b], [Prop A], [Mode A], [13]), where [b] i s a
mode of present at i on of the individual b, [Prop A] is a mode
of present at i on of the proposi t i on Prop A, [Mode A] is a
mode of present at i on of the mode of present at i on Mode A,
and [13] is a mode of present at i on of the belief rel at i on B.
So, on this account, the semantic cont ent of the sent ence c believes t hat
b believes t hat Fa with respect to a cont ext of ut t erance c and an
assignment funct i on f is ((c, ((b, (F*, a), (IF*]1, [ a ] l ) ) , B), {[bl2, [F*I 2, [a]2,
[ [ F * ] t ] 2 , [ [ a l ~ ] 2 , [ B ] 2 ) ) , B ) . (The numeri cal subscripts i ndi cat e the degree
of embeddi ng within at t i t ude constructions. Whet her [a]l is identical
with [a]2 - - and whet her [F*lt is identical with [F*I2 - - is a quest i on
which I shall not address here.)
It is a consequence of this account t hat the cont ri but i on which a t erm
makes to t he semantic cont ent of sentences in which it occurs is
dependent upon the or der expressions which occur in t hat sent ence - -
and, in particular, upon whet her or not it falls within the scope of any
proposi t i onal at t i t ude verbs. However, this is not to say t hat the seman-
SEMANTICS FOR ATTI TUDE ASCRIPTIONS 9
tics is not composi t i onal - - f or it is still t r ue t hat t he semant i c cont ent
of a sent ence is a st r uct ur ed ent i t y whi ch is const r uct ed f r om t he
semant i c cont ent s of t he expressi ons whi ch t oget her make up t he
sent ence. (Al t ernat i vel y, we coul d t hi nk of t he semant i c cont ent s of
expressi ons on par t i cul ar occasi ons of ut t er ance as infinite mat ri ces in
whi ch t her e appear appr opr i at e modes of pr esent at i on f or each degr ee
of embeddi ng in pr oposi t i onal at t i t ude ascriptions. However , on any
occasi on, onl y a finite "initial segment " of t he mat r i x will be r equi r ed in
or der t o det er mi ne t he semant i c cont ent of a sent ence in whi ch t hese
expressi ons occur. Of course, if we t ake this opt i on, we will not want t o
suppose t hat a compet ent speaker of t he l anguage must s omehow get
t hese infinite mat ri ces "i nt o her head"; rat her, t he i dea will be t hat a
compet ent speaker of t he l anguage is one who is able t o grasp t he
semant i c cont ent s of sent ences in a wi de vari et y of cont ext s of ut t er -
ance.)
v i i
So far, we have onl y t he bar est bones of a t heor y. The first meat on
t hese bones comes when we come t o l ook at what Kapl an woul d call
t he characters of t er ms - - i.e. when we come t o l ook at t hose funct i ons
f r om cont ext s of ut t er ance t o semant i c cont ent s whi ch char act er i se
anot her l evel of t he meani ngs of pr edi cat es and singular terms. For ,
while I woul d allow t hat a name like "Geor ge Bush" pi cks out t he ver y
same obj ect in a l arge r ange of cont ext s, I woul d also mai nt ai n t hat t he
mode of pr esent at i on whi ch is associ at ed wi t h this name will var y
across t hose same cont ext s.
Suppose t hat a per s on P ut t ers t he sent ence Fred believes that
George Bush is friendly i n a cont ext c. Suppose f ur t her that, in c, "Fr ed"
refers to Fr ed, and "Geor ge Bush" refers t o t he cur r ent U.S. presi dent ,
Geor ge Bush. The n P' s ut t er ance will have t he cont ent ((Fred, (is-
friendly*, George Bush}, ([is-friendly*], [George Bush]}}, B), wher e [is-
friendly*] and [George Bush] are modes of pr esent at i on whi ch are
suppl i ed f r om, or det er mi ned by, P' s cont ext of ut t er ance.
Thi s is still fairly thin, f or not hi ng has yet been said about t he nat ur e
of modes of pr esent at i on, nor about t he mechani sm by whi ch t hey get
suppl i ed f r om t he cont ext of ut t er ance of sent ences whi ch cont ai n
10 GRAHAM OPPY
propositional attitude verbs. (One important point to note is that I have
been using the term "mode of presentation" as a name for whatever it is
that it the additional component of the semantic content of proposi-
tional attitude ascriptions. There things may turn out to have nothing to
do with those other entities which have been called "modes of presenta-
tion" by other philosophers.)
VIII
On the question of the nature of these modes of presentation, I think
that it is natural to investigate the thought that modes of presentation
must be some sorts of constraints on the mental representations which
people have of those individuals, properties and functions which are
parts of the semantic contents of the expressions of the language in
question. Thus, in my example, it seems to me to be natural to suppose
that [George Bush] is a condition on the nature of mental representa-
tions of George Bush. Moreover, it seems to me to be plausible to
suggest that the proposition expressed by the sentence which P utters
will be true just in case Fred has mental representations of George
Bush and the property of being friendly which satisfy the constraints
imposed by [George Bush] and Iis-friendly*], and in virtue of which
representations it is correct to say that Fred attributes the property of
being friendly to George Bush.
This is still a rather schematic suggestion, since the notion of a
"condition on the nature of mental representations" obviously requires
further clarification. What sorts of conditions or constraints on the
nature of mental representations are candidates to be modes of presen-
tation?
One obvious thought is that a mode of presentation can be a restric-
tion on the sort of information which is encoded in a mental represen-
tation. Consider, for example, a case in which we are talking about the
astronomical exploits of the ancient Babylonians. When our uses of the
words "Hesperus" and "Phosphorus" have single embeddings in propo-
sitional attitude constructions, it seems reasonable to suppose that the
associated modes of presentation might be "object represented only as
visible in the morning sky" and "object represented only as visible in
the evening sky". Consequently, my utterance of the sentence The
SEMANTICS FOR ATTITUDE ASCRIPTIONS 11
ancient Babylonian astronomers believed that Hesperus is visible in the
morning sky has (subject to a certain amount of harmless simplifica-
tion) the semantic content ((the ancient Babylonian astronomers,
(Venus, is-visible-in-the-morning-sky*), (["object represented only as
visible in the morning sky"], [is-visible-in-the-morning-sky*])), B). More-
over, this utterance is true: for the ancient Babylonian astronomers did
stand in the belief relation to the proposition (Venus, is-visible-in-the-
morning-sky*) via a mental representation of Venus in which it was
only represented as being visible in the morning sky.
Similarly, my utterance of the sentence The ancient Babylonian
asuvnomers believed that Hesperus is Phosphorus has (subject to a cer-
tain amount of harmless simplification) the semantic content ((the an-
cient Babylonian astonomers, (Venus, Venus, is-identical-to*), (["object
represented only as visible in the morning sky"], ["object represented
only as visible in the evening sky"], [is-identical-to*])), B). Moreover,
this utterance is false: for the ancient Babylonian astonomers did not
stand in the belief relation to the propositions via two mental repre-
sentations of Venus, in one of which it was only represented as being
visible in the morning sky, and in the other of which it was only
represented as being visible in the evening sky.
One obvious question which now arises concerns my use of the
locution "is represented as". When I say that someone possesses a
mental representation of an object x in which that obj ect is represented
as having a propert y F, do I mean that the person in question must
have an explicit representation of the possession of the propert y F as
part of their representation of the propert y F? Not necessarily; for I
want to allow that someone may possess a mental representation of an
object x in which that object is represented as having the propert y F
provided that it is implicit in that person' s representation that x has the
propert y F. (Cummins (1987) discusses a number of different types of
what he calls "inexplicit content". Some of his suggestions seem to be
relevant (as does the discussion of "tacit belief" in Dennet t (1987) and
Richard (1990)); however, I do not propose to investigate here the
question of how we tell that it is implicit in a person' s representation of
x that x has the propert y F.)
Anot her obvious question is whether the above is an exhaustive
account of the nature of modes of presentation. Once again, the answer
12 GRAHAM OPPY
is negative: I don' t want to hol d t hat the onl y possible constraints on the
nat ure of ment al represent at i ons which coul d be relevant to the seman-
tic propoert i es of proposi t i onal at t i t ude ascriptions are constraints on
what t he things which are represent ed by t hose represent at i ons are
represent ed as -- for somet i mes constraints on the nat ur e of ment al
represent at i ons can (e.g.) concern the causal hi st ory of t hose repre-
sentations. (There is an illustration of this sort of case below.) However,
most of the interesting puzzl e cases in t he literature on proposi t i onal
at t i t ude ascriptions do involve cases in which the relevant modes of
present at i on are constraints on t he nat ure of ment al represent at i ons
which t ake t he f or m of constrains on what t he things which are repre-
sent ed by t hose represent at i ons are represent ed as.
There may be furt her possibilities. One of the reviewers of this paper
suggested t hat t here may well be "endogenous constraints on the ment al
structures themselves". In a sense, this is certainly true: human ment al
represent at i on is no doubt subject to endogenous constraints. However,
I doubt t hat our or di nar y practice of maki ng and using proposi t i onal
at t i t ude ascriptions is sensitive to such constraints. The reason why we
need modes of present at i on is t hat singular proposi t i ons do not always
suffice to pi ck out t he states of mi nd which are causally operat i ve in the
pr oduct i on of behaviour. But coul d it be t he case t hat the salient
di fference bet ween two ment al represent at i ons of a given singular
proposi t i on - - i.e. the di fference which we use to pi ck out the fact t hat
one rat her t han the ot her is causally involved in a given piece of
behavi our - - is t hat onl y one of t hem is subject to a certain sort of
endogenous const rai nt ?
Again, I shoul d emphasi se t hat modes of present at i on need to be
publicly identifiable and linguistically usabl e differences bet ween ment al
represent at i ons which have di fferent causal profiles, but which repre-
sent the same objects (or properties). Consequent l y, account s of
"modes of present at i on" such as t hat offered by McGi nn (1989) - - at
pp. 190- - 192 - - do not address my probl em. It seems very reasonabl e
to say t hat ment al represent at i ons must always be partial and aspect-
specific. Moreover, it also seems reasonabl e to say t hat this is an
expl anat i on of how it is t hat t here can be ment al represent at i ons which
have di fferent causal profiles, but which represent t he same objects (or
properties). But "modes of present at i on" in my sense can onl y be kinds
SEMANTICS FOR ATTITUDE ASCRIPTIONS 13
of ment al represent at i ons - - since t he ment al represent at i ons t hem-
selves will al most cert ai nl y be uni que to each person - - and it is
precisely the basis of t he syst em of classification into kinds which
remai ns to be explained. (Of course, I don' t mean to suggest t hat
McGi nn is confused about this: he makes it quite clear t hat he is talking
about ment al cont ent , not semant i c content. )
IX
What of t he quest i on of t he mechani sm by means of which modes of
present at i on get supplied f r om the cont ext of ut t erance of sentences
which cont ai n proposi t i onal at t i t ude verbs? Her e again, what I have to
suggest is rat her programmat i c. The mai n i nt ui t i on whi ch I have is t hat
t here are a variety of ways in which it coul d happen t hat a given mode
of present at i on of an object coul d become part i cul arl y salient in a
cont ext of ut t erance.
One possibility is t hat t he previ ous usage of a wor d in a conversat i on
can establish a connect i on to a condi t i on (or perhaps to a set of rel at ed
conditions) on ment al represent at i ons of objects and properties. Thus,
for example, consi der t he case in which Joe turns up to his first day of
work, and is i nt r oduced to his boss "Robert Gr ant " (a neat, clean,
sober-l ooki ng gent who sports a t hree piece suit and tie). Af t er work,
Joe goes to the pub, where he is regaled with stories about a hard-living
lunatic whom everyone refers to as "Bobby". Before long, Joe tires of
these stories, and so he says: "Enough stories about Bobby. Tell me
about my boss." Ther e is a somewhat pai ned silence, and t hen one
amongst the assembl ed crowd states the obvious fact: Joe hasn't realised
that Bobby is Robert Grant.
In this case, it seems to me to be reasonabl e to suggest t hat the prior
use of the word "Bobby" in the conversat i on serves to establish a
connect i on bet ween it and a mode of present at i on of the f or m "object
which possesses most of the propert i es ascri bed to "Bobby" duri ng the
precedi ng conversat i on". (Note, by the way, t hat the possibility of
const ruct i ng this sort of case provi des part of the mot i vat i on for
supposing t hat modes of present at i on are cont ext -dependent . Af t er all,
it seems clear t hat we coul d furt her stipulate that, in this example, it is
simply an acci dent t hat the name "Bobby" is the onl y one used in the
14 GRAHAM OPPY
conver sat i on in t he pub - - for, in fact, all of t he (ot her) part i ci pant s in
t he conver sat i on use t he t wo names mor e or less i nt erchangeabl y, at
least out si de of pr oposi t i onal at t i t ude const ruct i ons. I ndeed, exactly t he
same thing happened to Joe' s pr edeces s or in t he j ob, except t hat t he
t wo names wer e ever ywher e i nt erchanged!)
This suggestion may sound familiar; a r evi ewer of my paper sug-
gest ed t hat it is def ended by Evans ( 1982) and Bach (1987). I am not
sure about this, since it is not cl ear to me t hat Evans' account of nami ng
pract i ces tells us anyt hi ng about how names funct i on in pr oposi t i onal
at t i t ude ascriptions. (What Evans and Bach do descri be is ways in
whi ch it can come about t hat t he pat t er n of use of a name outside of
pr oposi t i onal at t i t ude ascri pt i ons can depend on "modes of pr esent a-
t i on of objects". But it is i mpor t ant to see t hat this is a di fferent subject
f r om t he one whi ch I discuss in my paper. ) I suspect t hat my suggestion
actually owes mor e t o Lewis (1979); however , I doubt t hat it is i mpor -
t ant to pur sue this suspicion.
Anot her possibility is t hat mutual knowl edge of t he interests and
pur poses of cert ai n part i ci pant s in a conver sat i on can serve t o establish
connect i ons bet ween cert ai n expressi ons and modes of present at i on.
Consi der, f or exampl e, Scot t Soames (1987)' s wel l -known exampl e
about Mar y and her nei ghbour.
Soames asks us t o suppose t hat Mary' s nei ghbour, Samuel Cl emens,
is in t he habi t of soliciting her opi ni on of his manuscri pt s bef or e he
sends t hem t o his publisher. Mor eover , he tells us t hat Mar y thinks
t hese manuscri pt s are wonderful , and hence thinks t hat Samuel Cl emens
( whom she knows onl y under t hat name) is a great writer. But, asks
Soames, does it fol l ow t hat she thinks t hat Mar k Twai n is a great
wri t er?
Well, first, Soames invites us to consi der t he following situation.
Mary, who is a st udent , has t aken a wri t t en exami nat i on; and her
t eacher explains t hat even t hough Mar y di d ver y well on t he exam, t he
r eason why she di dn' t get a per f ect scor e is that she di dn' t know t hat
Mar k Twai n is a great writer. In this cont ext , Soames suggests t hat t he
t eacher' s expl anat i on is perfect l y accept abl e.
However , Soames also invites us to consi der a di fferent conversa-
tion, whose pur pos e it to det er mi ne Mary' s opi ni on of vari ous authors.
The conversat i onal part i ci pant s, who use t he name "Mar k Twai n" t o
SEMANTICS FOR ATTITUDE ASCRIPTIONS 15
refer to Samuel Cl emens, want to know Mary' s opi ni on of him. Soames
suggests t hat one who knows Mary' s opi ni on could surely report that,
yes i ndeed, Mar y knows t hat Mar k Twai n is a great writer.
I suggest that, in the first case, the t eacher' s ut t erance of Mary didn't
know that Mark Twain is a great writer has (subject t o a certain amount
of harml ess simplification) the semant i c cont ent (Not ((Mary, (Mark
Twain, is-a-great-writer*), (["object represent ed as bearing the name
"Mar k Twain"], [is-a-great-writer])) K)). And I suggest that, in the
second case, the ut t erance of Mar y knows t hat Mar k Twai n is a great
writer has (subject to a certain amount of harmless simplification) the
semant i c cont ent ((Mary, (Mark Twain, is-a-great-writer*), (["object
represent ed as being the aut hor of such and such works[, [is-a-great-
writer])) K).
In these cases, it seems plausible to me to suggest t hat the modes of
present at i on are det er mi ned by mut ual knowl edge of t he interests and
purposes of the part i ci pant s in the conversations. In the first case, the
mai n i nt erest of t he part i ci pant s in the conversat i on is in the mar k
whi ch Mar y scored on a certain test. Consequent l y, their i nt erest in
Mary' s opi ni on of Mar k Twai n is an i nt erest in her opi ni on of hi m
under the represent at i ons of hi m which are rel evant to the test which
she has just t aken - - and so t hey are i nt erest ed in the propert i es which
she attributes to hi m under represent at i on of hi m in which she repre-
sents hi m as being called "Mar k Twain". Similarly, in the second case,
t he mai n i nt erest of the part i ci pant s in the conversat i on is the literary
merits of the aut hors of certain books. Consequent l y, their i nt erest in
Mary' s opi ni on of Mar k Twai n is an interest in her opi ni on of him as
the aut hor of certain books - - and so t hey are i nt erest ed in the proper-
ties whi ch she attributes to hi m under represent at i ons of hi m in which
she represent s hi m as the aut hor of certain books.
Finally, I don' t want to di scount the possibility t hat t here may be a
"defaul t setting" for the cont ri but i on of modes of present at i on to
proposi t i ons, viz: that, in the absence of defeating cont ext ual conditions,
the mode of present at i on which is associated with any t erm is simply
somet hi ng like "t he domi nant ment al represent at i on of [the object or
pr oper t y in question]" (where "domi nance" her e is cashed out in t erms
of the amount of i nformat i on which is cont ai ned in ment al represent a-
tions). In general, we are pret t y good at avoiding the sorts of situations
16 GRAHAM OPPY
in which the ancient Babyl oni an ast ronomers were placed with respect
to the planet Venus; for, in general, we t end to have one clearly domi-
nant concept i on of an object or pr oper t y - - and, moreover, in general,
it t ends to be this domi nant concept i on which is causally involved in
our dealings with t hose objects and properties. Consequent l y, it seems
to me to be reasonabl e to suggest t hat we onl y need to postulate a
mechani sm wher eby the defaul t presumpt i on that someone' s domi nant
concept i ons of objects and propert i es are involved in the pr oduct i on of
their states and behavi our can be overri dden. (Not e t hat this last
suggestion seems to give a mor e plausible way of handl i ng the case in
which Mar y is said to know that Mar k Twai n is a great author. Af t er
all, it is not clear t hat t here is anyt hi ng in the cont ext of ut t erance which
imposes constraints upon how Mar y thinks of Mar k Twain. We woul d
need to be told a lot mor e about the cont ext of ut t erance before a final
analysis of the case could be made.)
One criticism which might be levelled at the t heory which I have
sket ched is that the criticisms which have been levelled at neo-Fregean
theories in the last t went y years show that it is unt enabl e. It seems to
me t hat this criticism relies upon an i mport ant mi sunderst andi ng of
what it is t hat recent "ant i -Fregean" argument s have really established.
A useful way to approach this issue is to consi der a distinction which
has been drawn among various di fferent component s of the (allegedly)
Fregean not i on of sense. (Here, I follow Burge (1977) and Sal mon
(1981) (1986a).) Among the senses (or component s) of "sense" which
can be distinguished, t here are at least the following:
(i) Sensej: a purel y concept ual or totally descriptional repre-
sentation which all fully compet ent speakers associate with a
singular t erm
(ii) Sense:: a set or cluster of propert i es (represent ed in a dossier
or ment al file) which speakers (more or less) idiosyncrati-
cally associate with a singular t erm
(iii) Sense3: the mechani sm by which the reference of a singular
t erm is semantically det ermi ned
SEMANTICS FOR ATTI TUDE ASCRIPTIONS 17
(iv) Sense4: t he semant i c val ue of hyper i nt ensi onal occur r ences
of a singular t er m (i.e. of occur r ences of a singular t erm
whi ch fall within t he scope of verbs of pr oposi t i onal atti-
t ude)
Mor eover , it is useful t o not e t hat Frege' s own view - - or, at least, t he
view whi ch is most commonl y at t ri but ed to Frege, and whi ch is t aken to
be t he st andar d t arget of t heori st s such as Soames and Sal mon - - relies
on a not i on of sense whi ch is der i ved f r om t he i dent i fi cat i on of sense 1,
sense3, and sense 4 (or per haps sense2, sens%, and sense4).
Now, Frege' s own view, as thus charact eri sed, has been heavi l y
criticised - - especially, and most famousl y, by Saul Kr i pke (1980).
However , it is i mpor t ant to not e t hat t he most t hat this criticism seeks
to est abl i sh is t hat it is not t r ue t hat t her e is one not i on whi ch can
pl ausi bl y be i dent i fi ed with sense 1 + sense 3 + sense 4. Mor eover , it is
also wor t h not i ng t hat t he mai n ar gument of Naming And Necessity
aims t o show t hat it is a mi st ake to i dent i fy sense~ with sens% - - i.e. it
says not hi ng at all about sense 4.
Mor e recent l y, t her e have been t heori st s - - e.g. Sal mon, Soames - -
who have cont ended t hat t he onl y thing t hat sense4 can be is t he
r ef er ent of t he singular t er m in quest i on. However , as far as I can see,
t her e is ver y little posi t i ve ar gument whi ch has been gi ven f or this view;
rat her, t he mai n def ence of this view has been t hat it is har d t o see what
else sense 4 coul d be. (I have criticised t he posi t i ve ar gument s el sewher e
- - see "Why Semant i c I nnocence?", Australasian Journal of Philosophy,
f or t hcomi ng; consequent l y, I do not pr opos e to r epeat t hese criticisms
here. ) But one i mpor t ant resul t of t he above di scussi on is t hat t her e is
good r eason t o suppose t hat sense4's are modes of pr esent at i on in t he
sense t her e descri bed. Hence, i f - - as I have assert ed - - t her e are no
good posi t i ve argument s in f avour of t he view t hat t he onl y thing t hat
sense 4 can be is t he semant i c cont ent whi ch expressi ons have when t hey
occur out si de of hyper i nt ensi onal const ruct i ons, t hen t her e are no good
general obj ect i ons t o t he posi t i on whi ch I have def ended to be f ound in
t he existing l i t erat ure. ( Of course, as I argued earlier, t her e is t he
f ur t her poi nt t hat Sal mon and Soames need sense 4 in t hei r pragmat i cs;
consequent l y, t hey can har dl y claim t hat t hey have shown t hat neo-
Fr egean t heor i es ar e unt enabl e. )
18 GRAHAM OPPY
NOTE
* This paper is based on sections of my unpublished Ph.D. thesis: Attitude Problems
(Princeton, 1990). I would like to thank: Gil Harman, David Lewis, Scott Soames, Bas
van Fraassen, Richard Holton, Steve Rieber, Len Goddard, Allen Hazen, and Kai-Yee
Wong for their assistance with my thesis, and/or with this paper. Also, I thank the
editors and reviewers of Philosophical Studies for helpful comments and suggestions.
REFERENCES
Bach, K.: 1987, Thought And Reference Oxford: Oxford University Press.
Branquinho, J.: 1990, 'Are Salmon's Guises Disguised Fregean Senses?' Analysis.
Burge, T.: 1977, 'Belief De Re' Journal of Philosophy.
Cummins, R. : 1989, Meaning And Mental Represenmtion Cambridge: M.I.T. Press.
Dennett, D.: 1987, The InternationalStance Cambridge: M.I.T. Press.
Evans, G.: 1982, The Varieties of Reference Oxford: Oxford University Press.
Forbes, G.: 1987a, 'Indexieality And Intentionality: A Fregean Perspective', Philosophi-
cal Review.
Forbes, G.: 1987b, 'Review of Salmon (1986a)' Philosophical Review.
Kripke, S.: 1980, Naming and Necessity (2nd. edition) Harvard: Harvard University
Press.
Lewis, D.: 1979, ~ in a Language Game' Journal of Philosophical Logic.
McGinn, C.: 1989, Mental Content Oxford: Basil Blackwell.
Richard, M.: 1990, PropositionalAttitudes Cambridge: Cambridge University Press.
Salmon, N.: 1981, Reference and Essence Princeton: Princeton University Press.
Salmon, N.: 1986a, Frege's Puzzle Boston: M.I.T. Press.
Salmon, N.: 1986b, 'Reflexivity' Notre Dame Journal of Formal Logic; reprinted in
Salmon, N. and Soames, S. eds. (1988) Propositions And Attitudes Oxford: Oxford
University Press.
Salmon, N.; 1989, 'How to become a Millian Heir' Nous.
Soames, S.: 1987, 'Direct Reference, Propositional Attitudes, And Semantic Content'
Philosophical Topics; reprinted in Salmon, N. and Soames, S. eds. (1988) Proposi-
tions and Attitudes Oxford: Oxford University Press.
Soames, S.: 1988, "Direct Reference and Propositional Attitudes" Ms.
Depar t ment o f Phi l osophy
University of Wol l ongong
Wollongong, NS W 2500
Aust ral i a