Reference,Intentionality,and Purely Intentional Objects

Olav Asheim

Department of Philosophy,University of Oslo

Department of Philosophy,Nanjing University olav.asheim@ifikk.uio.no

Reference,Intentionality,and Purely Intentional Objects

Olav Asheim

Department of Philosophy,University of Oslo

Department of Philosophy,Nanjing University olav.asheim@ifikk.uio.no

.There are two big obstacles to making sense of quantification into belief and other propositional attitude contexts:one that the object referred to may not exist,the other that the principle of substitutivity of co-referential terms breaks down in such contexts.I shall argue in this paper that the first problem can be solved by recognizing objects of belief as entities that can be referred to regardless of whether they exist,and I shall sketch a theory of existentials that makes this plausible.In addition I shall argue that demonstrative reference is not in itself direct reference,and I shall argue that there is no problem of substitutivity when the co-referential terms in question are both used directly referentially.

1 Introduction

The ideas I am about to present in this paper come from an origin long ago.I started to think along these lines when I was working on my master thesis,which had as its topic Quine’s philosophy of language and mind.That was in 1968.I expressed my thoughts in a more mature form in my doctoral dissertation Reference and Intentionality,which was published as a book by Solum Forlag in 1992.1See also[1]on the same topic.The book has later been translated to Chinese,and was published by Nanjing University Press in 2014.

My ideas are a reaction to the problems that arise when we try to make sense of quantification into belief and other attitude contexts,and meant to be a solution to them.Part of my solution,and the most important part,is to recognize purely intentional objects—objects that can be identified only by their intentional properties,by what people feel and think about them and how they relate to them intentionally—as entities that can be referred to and quantified over.

Our point of departure is the intuition that tells us that some beliefs,but not all, are about objects directly.One example is Paul’s belief about his colleague Joan that she is a single woman.This belief seems to be different in logical form from the belief Paul also has that the shortest spy in the world is a single woman.The latterbelief does not have the implication that there is some particular object of which Paul believes that it is a single woman,a spy,and shorter than any other spy,but his belief about his colleague Joan that she is a single woman has the implication that there is a particular object of which Paul believes that it is a single woman,if our intuition can be trusted.This kind of belief is often called“belief de re”(of the object)to distinguish it from belief that is merely de dicto(of what is said).

A belief is said to be de re insofar as it allows the inference from“s believes that Fa”,where“a”is some singular constant,to“∃x(s believes that Fx)”,quantifying into the belief context.This inference can only be valid if the singular constant behaves logically like a variable that has been assigned an object as its value,so that existential generalization can be meaningfully performed on it.If a belief context does not contain a singular constant that exhibits the logical behavior of a variable, the belief is only de dicto,and the belief context cannot be quantified into.

Ignoring the reference to the bicycle,the beliefs reported with(1)and(2)below seem to be clearly different in logical form,(1)de dicto,and(2)de re.

(1)Peter believes that his bicycle has been stolen.(de dicto)

(2)Peter believes that Richard has stolen his bicycle.(de re)

When your bicycle is stolen,it makes a difference that you have a suspect.

The difference between beliefs which are about objects de re and beliefs which are not about objects in this way seems to be vast.Phenomenologically,our always changing environment typically contains some known objects,some perceived objects,and many unknown and unperceived objects which are even so present.Perceived objects may be known or unknown.If there is no distinction between known and unknown,it will be difficult to find one’s way around.2Michael([14])objectstothisthattherearemoredistinctionsthanonebetweenknownandunknown, and that none of these distinctions sustains quantification into attitude contexts.He argues that the obstacles there are to quantifying into such contexts are insurmountable.In this paper I shall show that they can be overcome.This is what our intuition tells us.

But maybe we should not trust our intuition here.For there are two apparently insurmountable obstacles to making sense of quantification into belief contexts,and hence to making sense of belief de re as a special kind of belief(and not only as a way of reporting beliefs).

One obstacle is that the principle of substitutivity of coreferential terms seems to break down in such context,as was pointed out by Quine in his seminal paper[18] with an instructive example:First,there is“a certain man in brown hat whom Ralph has glimpsed several times under questionable circumstances”.Second,there is“a gray-haired man,vaguely known to Ralph as rather a pillar of community,whom Ralph is not aware of having seen except once at the beach.”The man’s name is“Bernard J.Ortcutt”.Now Ralph believes that the man in brown hat is a spy,buthe does not believe that Bernard J.Ortcutt is a spy.But,unbeknownst to Ralph,the man in brown hat is in fact Bernard J.Ortcutt.How can we say,then,that there is something(somebody)Ralph believes de re of that it is a spy?Who could that possibly be?

The second obstacle to making sense of quantifying in is that the object an attitude like a belief is directed at,may not exist.Some call this“the problem of possible reference failure”because they think that an object that does not exist cannot be referred to.One example of an object that does not exist,but even so seems to be the object of an attitude,is Santa Claus:Danny believes that Santa Claus will bring him gifts on Christmas Eve.He also believes that he has met Santa Claus in person.But Santa Claus does not exist.Another example:A businessman thought he had one secretary,but in reality there were two twin sisters who shared the job between them. They were so similar that the man never detected the difference,and he only saw one at a time as long as the illusion lasted.The person he believed was his secretary did not exist;there was no such person.

2 The Problem of Possible Reference Failure

In this paper I shall first show how the second obstacle can be overcome.That will make it easier to get rid of the first obstacle afterwards.

A viable solution to the problem of possible reference failure should not obscure the de re/de dicto distinction,and it should respect first-person authority on the question whether a belief is expressed or not and what the belief may be,to avoid serious complications.3This is a different question from that of what has been said.It is not so easy to find a solution that meets these requirements. Transforming“think”and“believe”into success verbs,as Evans([5])proposed,suggesting the expression“essaying a thought”,so that you can be said to fail in thinking something you feel you think,does not respect first-person authority,and it is also a threat to the de re/de dicto distinction since it seems that one can only fail in thinking thoughts about particulars.As long as the thought content can be expressed by a quantified sentence,and no irreducibly singular terms are involved,it seems that a thought attempt will always be successful.And there is usually more than one quantified sentence that corresponds loosely in meaning to a sentence that contains an alleged irreducibly singular term.In some contexts the singular term can be replaced by a description of the name’s bearer,but not in belief contexts.The upshot is that belief contexts cannot be quantified into.

2.1 ‘The easy way out’

There is an obvious solution to the problem of possible reference failure,which is also the only viable solution I can see,and that is to take the easy way out,and recognize objects intentional acts and attitudes are directed at as objects we can refer toandquantifyoverregardlessofwhethertheyexist.Therewillnolongerbeadanger of a reference failure,and hence no problem.Note that taking this way out does not force us to recognize every entity that is said to be non-existent as a(potential)value of quantificational variables.Recognizing only entities that are objects of intentional acts and attitudes as objects we can refer to and quantify over even when they do not exist,is enough to solve the problem.

Meinong is notorious for having proposed a solution that consists in recognizing non-existing objects.([13])I think his bad reputation has its roots in the greatly exaggerated accusations Bertrand Russell made against him in the paper[21].Russell made similar unreasonable accusations against Frege concerning the latter’s concept of sense.In reality,it was his own developments of Meinong’s and Frege’s theories Russell attacked in his paper.If you believe what Russell says there,you will come to believe that Meinong’s theory is absurd,as many have been taught to think.But Meinong’s theory is not absurd,and it is my impression that more and more philosophers understand that today.Some,among them notably Priest([16]),have taken Meinongianismfurtheranddevelopedveryinterestingtheories.Evenso,themajority ofphilosophersofmindandlanguagewillprobablystillrejectatheoryofnon-existent objects almost instinctively.And perhaps not without a reason,for Meinong’s theory has indeed its weak points.

On Meinongian theories,the predicate of existence is an almost normal predicate,expressing an almost ordinary property,in that not only existence,but also negated existence can be combined with most other properties so that we can talk about non-existing people,for instance.4The theories of Crane([3,4])and of Moltmann([15])should not be labeled Meinongian,I think, though both authors maintain that there are objects that do not exist,and treat existence as a predicate. They distinguish themselves from Meinongians by denying that non-existence can ever be combined with other properties.This leads to hard problems Meinongians have to solve,notably the problem of contradictory objects,among them not only the round square,but also the round square that exists.Besides,when the predicate of existence is taken as a primitive,the connection between this predicate and the existential quantifier is broken.It is of importance here to be aware that not only theories that recognize non-existent objects as objects of reference,but also theories that recognize only objects that exist,must recognize a predicate of existence.Insofar as“X exists”makes sense,“exists”must be a predicate.If you hold that only existent objects can be quantified over,then the predicate of existence expressed as“∃x(x=_)”,which is true of everything,should do for you.This predicate of exis-tence is closely connected with the existential quantifier,and in a sense reducible to it.In contrast to this,Meinongian theories see the predicate of existence as unrelated to what is called“the existential quantifier”by most,but“the particular quantifier”by noneists like Priest.5The term“noneism”for this point of view was coined by R.Routley in[20].Noneism makes a clear distinction between“there are white unicorns”and“some unicorns are white”,claiming that asserting the latter instead of the former carries no ontological commitments whatsoever.

There can be no doubt that words like“exist”and“be”are ambiguous,and that there are many concepts of being and existence.But I think even so that the word“exist”is polysemic rather than homonymous,so that the different senses of the word are related in meaning to each other and not completely separate.It speaks against Meinongianism that it cannot account for the relationship between the predicate of existence and the existential quantifier.

2.2 Purely intentional objects

ThetheoryIshallarguefordoesnothavetheundesirableconsequencesofMeinongian theories.On my theory,negated existence cannot be combined with positive properties of material objects,so there are no non-existent people;and there is a clear connection between the existential quantifier and the predicate of existence,as we shall see later.But we need some preparation for that,so let me first introduce some important concepts.

By an intentional object I mean something that is the object of some intentional act or attitude;not necessarily the act or attitude of an individual.Since there are different kinds of intentional acts and attitudes,there are also different kinds of intentional objects.As an important special case I call objects beliefs are directed at“doxastic objects”.Many philosophers use the term“intentional object”in a different sense,to mean an intermediary between an intentional act or attitude and the real object it is directed at if there is one.This concept of an intentional object must be clearly distinguished from my concept;in my terminology the intentional object is identical with the real object the act or attitude is directed at if there is one,and I see no need for an intermediary entity.If there is no‘real’object,say a person,the intentional act or attitude is directed at,then the intentional object is purely intentional, and if the attitude is a(de re)belief,then its doxastic object is purely doxastic.

In my book Reference and Intentionality I only talk about intentional and purely intentional objects,and I do not introduce the concept of doxastic objects as a special case.But today I see it as important to distinguish between purely intentional objects in general,and purely doxastic objects in particular.For if Ingarden([8,9])is right, all artifacts are purely intentional objects.Ingarden argues that for instance a church is not identical to the‘heap of stones’that makes up its material substratum.Thesestones may be exchanged one by one until none of the original ones remain without affecting the identity of the church as such.The church may even be destroyed completely by a bomb,say,and then rebuilt,and we would still talk about it as the same church.The church remains,and it remains the same,as long we regard it as this particular church.It is the intentional properties of the church that define it,not its physical properties.If the church is desecrated and the building is put to a new use, say as a museum,then the church is no longer there,only the structure that used to be its material substratum.

This theory is controversial.But as long as it is not clear that artifacts are not purely intentional objects,I find it necessary to distinguish between intentional object in general and doxastic objects as a special case.For purely doxastic objects are always said not to exist,whereas artifacts are said to exist,whether they be purely intentional objects or not.Also purely fictional objects,like characters in a novel,are said not to exist,and they,too,are counted as purely intentional by Ingarden.

By an intentional predicate I mean a predicate that has the structure“s ints that F_”,where“s”stands for a subject(not necessarily an individual),“ints”for some intentional verb construction(like“wishes”or“imagines”),and“F”for a predicate; or is a predicate that supervenes(in a loose sense)on other intentional predicates, like“_is popular”,which presupposes for its truth about a particular object that many people have positive attitudes to it.Intentional predicates express intentional properties.Note that it makes sense to talk about intentional predicates if and only if quantification into intentional contexts makes sense:these are two sides of the same coin.

A predicate that has the structure“s believes that F_”I call a doxastic predicate.It is clear from the definition of“intentional predicate”that doxastic predicates are intentional predicates.We can also say that doxastic predicates express doxastic properties.

Nowmoreprecisedefinitionsof“intentionalobject”,“purelyintentionalobject”,“doxastic object”,and“purely doxastic object”can be given.An intentional object is an object some intentional predicate is true of(or:an object that has intentional properties).A doxastic object,as a special case,is an object some doxastic predicate is true of(or:an object that has doxastic properties).

Anobjectispurelyintentionalifandonlyifitisindividuatedbyintentionalproperties alone.Insofar as Ingarden is right,and artifacts are purely intentional objects, these objects have material properties in addition to their intentional ones,but their material properties do not suffice to give them an identity as the artifacts they are.

Finally,an object is purely doxastic if and only if it is individuated by some doxastic property,but not by any substantial non-intentional property.This holds true of SantaClaus,forexample,andalsoofthenon-existentobjectthebusinessmanthought was his secretary.He believed things about that object that he did not believe about any other,but his beliefs were fundamentally mistaken:there was no such person.

By saying that a predicate is individuating,I mean that the predicate is true of one and only one object if of anything.When I say that an object is individuated by a certain property,I mean that the individuating predicate this property is expressed by,is true of the object,and can be used to identify it.By saying that a predicate P is constitutiveofanobjectaforasubjects(whichmaybeanindividualoracommunity), I mean that P is an individuating predicate s uses for re-identification of a:if P is true of x,then x=a,and if P is not true of a,then P is true of nothing.Moreover, if P is not true of the object a it is constitutive of,and a is believed by s to be a dog, for example,then a is not a dog,not a living being,not anything natural or physical, but a purely doxastic object.

In order for someone,s,to believe something about an object a directly in the first place,there must be some predicate,P,that is constitutive of a for s,and s must believe that P is true of a;s must have some way of identifying a as the object her belief is about.You cannot be said to believe something about an object if you don’t know which object you hold the belief about.A subject who gives us nothing but a name,and is not able to indicate,be it by description,demonstration,or in any other way,which object she refers to by using this name,cannot be taken seriously;there is no reason to think that there really is an object she believes something about.Now Kripke([11])argues that we can succeed in referring to something by name without havinganyindividuatinginformationaboutthename’sbearer.Forinstance,someone who only knows that Cicero was a Roman orator,which is not enough to individuate him,can even so refer to Cicero and say things about him.Whether this person expressesabeliefderebysayingthatCicerowasanorator,isadifferentquestion.Ishall discussthatlater.However,IthinkKripkeiswrong,forthereactuallyisadefinitedescription the speaker can use to identify Cicero:she can say“the Roman orator called‘Cicero’”.This works as long as there was only one Roman orator named“Cicero”. Had there been two or more,or none at all,it would not under the circumstances be possible to decide whom the speaker refers to by that name.The reason Kripke is not willing to consider this definite description is that he regards it as circular.But there is really no vicious circle in it,not as long as there are other people who can identify Cicero without quoting his name in a description of him.

Doxastic objects will always be individuated by doxastic properties,whether or not they are also individuated by non-intentional properties.This is easy to see now in light of the above.Given that P is an individuating predicate,also“s believes that P”will be an individuating predicate.As long as s is rational,she will not believe that an individuating predicate is true of more than one object.We can be sure,then, that even when P is false of a,“s believes that Pa will be true,giving a an identity as a purely doxastic object.

For a subject,s,to believe something about an object directly,she must believe somethingaboutthatobjectthatshedoes notbelieveaboutanyother;shemusthavea wayofkeepingthedifferentobjectssheholdsderebeliefsaboutapart.Thisisatleastthe way it has to be in a consistent model of a belief system.Let me use the resources of quantified modal logic to show that.The‘possible worlds’that interest us here,are ways the world could be to be compatible with the actual beliefs of a subject,s;I will follow the custom of talking about them as doxastic alternatives for s.The point is that in every doxastic alternative for s,everything s believes is true.There is a small part of the world s thinks she knows,and there is a much larger part of the world that is unknown to her,but even so known to be there.All doxastic alternatives for s share a common core,and differ in that which is not known to the subject.

In each doxastic alternative there is a maybe infinite number of objects of different kinds.Now,for a belief to be de re,it must be true of the same object in every doxastic alternative.The objects such beliefs are about belong to the part of the world that is known to the subject,and these objects are the ones the subject is acquainted with.If the belief concerns a particular object,but is only de dicto,it will typically be true of different objects in different doxastic alternatives.Let a be an object s holds de re beliefs about.Then in every doxastic alternative for s,the same set of predicates are true of a.But these predicates are not necessarily true of a in the real world.Let us take as an example the purely intentional object the businessman thought was his secretary;we can call her“Marianne”,laying down in addition that the twin sisters who shared the job between them were called respectively“Mary”and“Anne”.He would of course hold a lot of beliefs about Marianne,some of them uniquely about her,and one of those(or a cluster of them)even constitutive of her for him,and everything he believes about Marianne is true of her in every doxastic alternative for him.But not in the real world,where the vast majority of those beliefs are false,in particular the beliefs that are constitutive of her,since Marianne does not exist and is not a person.What even so gives Marianne an identity in the real world,distinguishing her from every other object,are the beliefs the businessman holds about her and only about her.

In this case,there is one purely doxastic object,Marianne,and two real objects, Mary and Anne.The businessman stands in an intentional relation to Marianne,and in different causal relations to Mary and Anne.In the case of Danny,who believes in Santa Claus,there is one purely doxastic object,Santa Claus,and one object that is notpurelydoxastic,namelyDanny’sunclewhodressedupasSantaonChristmasEve last year.Danny relates intentionally to Santa Claus and to his uncle as two different person,but he relates causally only to his uncle,also when he relates intentionally to Santa Claus.In a case where a purely doxastic object has its origin in a hallucination or some other kind of delusion,there may be no object the subject stands in a relevant causal relation to at all,only an intentional relation to something that does not exist.

2.3 The problem of non-existence

This brings up a new question:what does it mean that a purely doxastic object does not exist?We have the material for an answer to this question already.I have argued that for a subject to believe something about an object directly,the subject must believe something about it that is constitutive of it,something that can be true only of that object.The short answer is that a doxastic object,a,exists if and only if every predicate,P,that is constitutive of it for a subject,is true of it.If P is not true of a,then a is a purely doxastic object.

When Kripke([11])argues that no property determines the reference of a proper name,he seems to be limiting his notion of properties to purely qualitative intrinsic ones.In my opinion,the question whether Aristotle has existed depends more on properties that are neither qualitative nor intrinsic,but are instead properties which have to do with the etiology and history of“Aristotle”as the name of a certain famous philosopher,than it depends on whether Aristotle really was a great philosopher,the teacher of Alexander,a student of Plato’s,and so on.When a name is used as a name of something that is supposed to have physical properties,we regularly assume there to be a causal connection between the name as used in accordance with a certain convention,and the object this convention lets us refer to with it.The property of being the object(a precursor of)a certain proper name was given to initially in a tradition of using the name to refer to that object leading up to present usage,is of central importance for the question which object the name refers to.If there is no causal connection,and there ought to be one,the referent of the name will be said to be non-existent,and it can have no(positive)physical properties.(A different account must be given for reference to abstract objects.)Kripke should not disagree;this comes close to his own account of proper names.

There will always be more than one predicate that is constitutive of an object for a subject,and I think that different constitutive predicates in some cases express distinctly different properties too.Take as an example the beliefs Bob has of Chris, a close friend of his who is also a colleague he has worked together with for years. Some of these beliefs fall in the category‘Chris as a friend’,some in the category‘Chris as a colleague’;the two categories do of course overlap.From each category separately it is possible to extract and construct constitutive predicates.Constitutive predicates that belong to the same category may not express very different properties, but when the constitutive predicates belong to different categories,one being a‘Chris as a friend’predicate and the other a‘Chris as a colleague’predicate,the properties they express can be different enough.Even so,if Chris is in lack of one of these properties,he is also in lack of the other one,and he is not a real person.

Nothing prevents a predicate that is constitutive of a certain object for a certain subject from being constitutive of the same object for another subject as well.That is sowhetherthepredicateistrueoftheobjectitisconstitutiveofornot,asisclearfromexamples like that of Santa Claus.6Gods are perhaps better examples.It makes sense to ask whether the Christian god is the same as the Muslim god independently of the question whether these gods exist.Some of the de re beliefs a subject holds about an object may be private,but in most cases many of them are beliefs that are shared with others.It is then reasonable to assume that in the linguistic community that consists of the users of a certain proper name N to refer to a certain object,there will be at least one predicate P that is constitutive of the referent of N for the community as a whole.

My view of proper names and predicates that are constitutive of their referents may seem to come close to a descriptivist treatment of proper names,especially to the‘cluster theory’of Searle([24]),who thinks there is not one property that determines the reference of a proper name,but a cluster of them,some more and some less important.But there is an important difference.On descriptivist theories,proper names can be eliminated in favor of definite descriptions based on what I call“constitutive predicates”.Let“CAristotle”stand for a predicate that is constitutive of the referent of the name“Aristotle”for a speaker or a linguistic community.If descriptivism is right,we can replace“Aristotle”in the statement“Aristotle rejected the Platonic theory of forms”with“the object,x,such that CAristotlex”.Following Quine([17]),we can say this more simply:we can introduce“aristotelizes”to replace the construction“is Aristotle”seen as a unit,and talk about the aristotelizer.Now we can everywhere replace the proper name“Aristotle”with the definite description“the aristotelizer”, and“Aristotle rejected the Platonic theory of forms”will become“the aristotelizer rejected the Platonic theory of forms”.The description“the aristotelizer”can then be eliminated along Russell’s lines.

In the general case,when α is a proper name,and P is a predicate that is constitutive of the referent of α for a speaker or a linguistic community,the definite description“the object P is true of”can,on descriptivism,replace α in any predicative statement“Fα”,and the description can then be eliminated to yield“there is one an only one object,x,such that P is true of x&Fx”.On the theory I am arguing for,that reduction is impossible.Let us use“Marianne”,the name of the businessman’s non-existing secretary,once more as our example,and let“CMarianne”stand for a predicate that is constitutive of the referent of“Marianne”for the businessman. Now it is an important point that every object that can be referred to and quantified over must have an identity,and I have argued that purely doxastic objects do have identities so that we can quantify over them.It should be clear,then,that the statement“Marianne is self-identical”is true.But the statement“the object,x,such that CMariannex is self-identical”is false,because there is no such object.We see the falsity clearly when we eliminate the description contextually,so that we get“there is one and only one object,x,such that CMariannex&x is self-identical”.

In my terminology,a directly referential term is a(singular)term that when usedin the utterance of a predicative statement“Fa”,picks out the object the statement is about,represented by“a”,but makes no contribution to its propositional content, represented by“F”.A directly referential term behaves logically as a variable that has been assigned an object as its value.But there is is no law(of logic)that forbids directly referential terms to be associated with reference-determining conditions,expressed by constitutive predicates;instead there seems to be a need for such conditions.I have accounted for the difference between“Fa”and“∃x(Cax&Fx)”by an example above.

On my theory,when“a”is a directly referential term,“a exists”is true if and only if“∃xCax”is true,if and only if Caa.As long as we are concerned only with singularexistentials,thisseemstobeadequateasananalysis,andthiswastheaccount I gave of them in my book Reference and Intentionality.The merit of the account is that the predicate of existence is not taken as a primitive,and the existential quantifier is allowed to play an important part.And why should not a proper name behave as a referentialterminpredicativestatements,andevensobehaveasa(constitutive)predicate in an existential?But the weakness of this account is that its scope is so narrow: although it seems to give the right truth conditions for affirmed and negated singular existentials like“Vladimir Putin exists”and“Santa Claus does not exist”,it cannot account for plural existentials like“Some Biblical characters existed,while others did not exist”.7This example is borrowed from T.Crane,used by him repeatedly([3,4]).Therefore I will here extend my account by proposing the following analysis of the existential“x exists”,where“x”is a variable:The open sentence“x exists”is true of an object z if and only if for every predicate,P,if P is constitutive of z for some subject,then“∃xPx”is true.(As constitutive for z,P can only be true of z.)This definition is built on the account given for singular existentials already. Non-intentionalobjects existbydefault,sincetheyhavenoconstitutivepropertiesfor anyone.

3 The Problem of Substitutivity Failure

I take myself to have shown by now that recognizing purely doxastic objects as denizensofone’suniverseofdiscourseisonewayofsolving,orratherdissolving,the problem of possible reference failure.But this is not enough for us to conclude that quantifying into belief and other attitude contexts makes sense,for there is still the problem of substitutivity failure:the problem it is for quantification into belief contexts to make sense that co-referential singular terms are not as a rule interchangeable in attitude contexts.

3.1 Demonstratives and direct reference

It is easily seen that this is only a problem if the co-referential terms that are not interchangeable are both directly referential in the sense that they contribute nothing to the predicative content of a statement.Insofar as they make a contribution to the predicative content of a statement,they are subject to the fact that belief and other attitude reports are unredeemably extensionally opaque:you may believe one thing about animals with a heart,and something else about animals with kidneys,even though every animal with a heart is also an animal with kidneys,and the other way round.Most definite descriptions are probably not directly referential,and because of that they should not worry us.But demonstrative descriptions still have to be considered,sincemostproponentsofdirectreferencetheorythinkthatnotonlyproper names,but also demonstratives and other indexicals are(typically used as)directly referring terms.I do not agree.To see demonstratives as directly referring terms is to ask for trouble.This is also the point of Eubulides’s Masked Man:“You don’t know the masked man?The masked man is your father.This means you don’t know your father.”8See D.Laertius,Vita philosophorum,II,108.This puzzling argument is commonly analyzed and dissolved as a sophism, a quaternio terminorum,“fourfolding of terms”,where one term counts for two in a three-term syllogism.The word“know”is used in two different senses,first in the senseofrecognize,andtheninthesenseofbeingacquaintedwith.Buttoatheorythat treats both demonstratives and proper names as directly referring terms,the Masked Man is a real challenge,as we shall see.([25])

What you refer to as“this”may have no clear identity to you apart from being the object pointed at,or in some other way the salient object in the contextual environment.That is why it makes sense to ask“what is this?”and“who is this?”.You see a person you cannot recognize cross the road in front of the car you are driving, and there is no way you can know that it is not your father,say.In spite of this,many have argued that indexicals are directly referential terms,trying to explain away the trouble this view leads to.It is easy to explain why it is so tempting to see demonstratives as directly referring terms,even as a paradigm case.For it is only natural to regard demonstrative reference as the most direct reference there can be:pointing directlyatanobject.Buttheabilitytopointatanobjectperceivedandcallingit“this”is not necessarily a sign of acquaintance with it.I think Russell was right in thinking that there is a close connection between knowledge by acquaintance as different from knowledge by description and direct reference([21]),while I think that he was wrong in treating“this”as a directly referential term([22]).Because of the problems it leads to,we ought to resist the temptation to see indexicals as directly referring terms on a par with proper names as long as there is an alternative way of analyzing them as functional expressions.

In my book Reference and Intentionality I give an analysis of indexicals,in-cluding demonstrative descriptions,based on Hans Reichenbach’s theory of tokenreflexivity.According to Reichenbach,indexicals are“descriptions in which the individual referred to is the act of speaking.We have special words to indicate this reference;such words are‘I’,‘you’,‘here’,‘now’,‘this’.…It is easily seen that all these words can be defined in terms of the phrase‘this token’.The word‘I’,for instance,means the same as‘the person who utters this token’;‘now’means the same as‘the time at which this token is uttered’;‘this table’means the same as‘the table pointed to by a gesture accompanying this token’.”([19],p.284)An important consequence of his theory is that indexicals are not directly referential terms.

My view of indexicals has undergone a change since I wrote the book.I still see these terms as reflexive in a way,but not as token-reflexive in Reichenbach’s sense. Today I prefer to call indexicals“context-reflexive”.The reason for this change is, first,that it is not so clear what Reichenbach means by“token”:He starts his brief discussion by talking about the token as the act of speaking.Here token-reflexivity means utterance-reflexivity.However,towards the end of his discussion,he takes“token”to mean the particular copy of a text.This is an important difference.Taking“token”to mean“particular copy”leads to difficulties,cf.Reichenbach’s own example:“Although the sentence‘the page on which this sentence is written has the number287’istrueforeverycopyofthisbookitsmeaningisdifferentforeverycopy; it refers to the page of the individual copy in which it is printed.”([19],p.287)It is not easy to accept this.Second,even if we take“token”to mean“utterance”,several problems remain.I will just mention one:on Reichenbachs theory it seems to be impossible to repeat a statement that contains an indexical element,as most statements do.Therefore I have come to think that these terms should better be called“contextreflexive”.My present understanding of indexicals is a view of them as functionals expressions with the context of the statement(request,question)made with an utterance of them as their unarticulated argument.([2])They could have been called statement-reflexive were it not for the fact that indexicals are used in other kinds of speech act too.However,this change in my view of indexicals has only minor and mostly negligible consequences for my argument in the book.

I will conclude my discussion here by saying that if we take the typical use of indexicals to be directly referential in the sense that they contribute nothing to the propositional content of the statements they help express by picking out their referents,and at the same time let non-indexical terms be directly referential too,this will necessarily lead to substitutivity problems.There is for instance no reason to expect thataco-referentialtermcanreplaceademonstrativedescriptioninanattitudecontext without affecting the truth-value of the report,which is clearly shown by Eubulides’s Masked Man.

BeforeIcontinue,Iwanttoemphasizethatthoughthereprobablyisatypicaluse of proper names,a typical use of indexicals,and a typical use of definite descriptions, this does not mean that every case of using a certain singular term is a case of usingit the way that is typical of that kind of term.As I see it,demonstratives and other indexicals may on occasion be used to refer directly,and the same is true of definite descriptions.On the other hand,the occurrence of a proper name in a statement does not mean that the name is used to refer directly there,even though that seems to be the typical use of proper names.Kripke([11])has argued convincingly that proper names are(typically)rigid designators,and his argument goes a long way towards establishing that they are directly referential terms as well.Even so,many examples show that co-referential proper names cannot always be substituted for each other in attitude contexts salva veritate.One often cited example is that of John,who believes that Cicero was a great orator,but not that Tully was a great orator,in spite of the fact that“Cicero”and“Tully”are two names of the same man.This is the real substitutivity problem,and the challenge we have to meet.

3.2 Speaker’s reference and semantic reference

We can talk about semantic meaning and semantic reference,and we can also talk about speaker’s meaning and speaker’s reference.(See Kripke[12].)Semantic meaning and semantic reference are determined by and for a linguistic community as a whole,while speaker’s meaning and speaker’s reference are determined by the speaker’s linguistic understanding.Semantic reference and speaker’s reference may coincide,but they can also come apart,and as I see it,there are two ways they can do that.The speaker may refer to an object that is not the semantic referent of the name,or the speaker may refer to the semantic referent,but in a non-standard way. For instance,when John refers to Tully,his reference is not direct;instead he refers to Tully as“the person I have heard being referred to as‘Tully’”.As long as there is only one such person,the description is good enough,and if that person is Cicero, speaker’sreferentisthesemanticreferent.Iftheobject(person)thespeakerhasheard being referred to as‘Tully’is somebody else than Cicero,then not only speaker’s reference deviates from semantic reference,but speaker’s referent is also different from semantic referent.

For the question of what a speaker believes,it is speaker’s reference,not semantic reference,that is decisive.It is the speaker’s understanding of the words used that is the key to her belief,not their normal communal use.Say that Anna has gotten the names of Jimmy Carter and John F.Kennedy mixed up to the effect that she refers to Carter as“Kennedy”and to Kennedy as“Carter”.When she assents to the sentence“Kennedy won the Nobel peace prize”,she means to assert that Carter won the prize; that is what she believes.And when she asserts that Carter was murdered in Dallas, she expresses a belief about Kennedy,not about Carter.Even so,it is a fact of linguistic usage that the latter belief could be reported as“Anna believes that Kennedy won the Nobel peace prize”,and this report will also be true when interpreted as“Anna believes that the person she calls‘Kennedy’won the Nobel peace prize”,taking thename“Kennedy”as being the speaker’s term with its speaker’s reference.On another interpretation the name“Kennedy”is the reporter’s term,speaker’s reference coinciding with semantic reference,we shall presume.Then the belief report is simply false.The difference between these two ways of using a term in a belief report,either as the speaker’s term or as the reporter’s,is most important for the truth or falsity of the report,but seldom emphasized.Also,there is no way of marking this difference grammatically in any language I know,but there may even so be languages where it is marked grammatically,for it is clear that it is possible to mark it.I will do that now by using the subscripts“S”for“speaker”and“R”for reporter.Thus“Anna believes thatKennedySwontheNobelpeaceprize”istrue,and“AnnabelievesthatKennedyRwon the Nobel peace prize”is false.

3.3 Parasitic usage and the principle of disquotation

Kripke does not take the difference between speaker’s reference and semantic reference fully into account in his formulation of the principle of disquotation:“If a normalEnglishspeaker,onreflection,sincerelyassentsto‘p’,thenhebelievesthatp.”([10],pp.248–249;his italics.)As Kripke uses the phrase,a normal English speaker isonewho“usesallwordsinthesentenceinastandardway,combinesthemaccording to the appropriate syntax,etc.:in short,he uses the sentence to mean what a normal speaker should mean by it”.This is rather vague,but it seems at least to be clear that by this criterion Anna is not a normal English speaker.But John in our example above should satisfy the criterion.That is clear from Kripke’s use of examples.If a speaker uses a term to refer to an object that is not the term’s semantic referent,then he is not a normal speaker,but as long as speaker’s referent is the semantic referent, Kripke’s criterion does not distinguish between cases where speaker’s reference(and not only the referent)coincides with semantic reference,and cases in which speaker’s reference differs from semantic reference.But then Kripke’s principle is not valid as it is stated.We cannot expect a principle of disquotation to hold if it is not restricted to apply only to speakers who are fully competent users of every word in the sentence uttered,speaker’s reference coinciding completely with semantic reference,both in referent and in the way it is picked out.Let me show this by an example.

Imagine Bill,a normal speaker of English,sincerely and on reflection assenting to“carbon dioxide is an element”.Should we really take Bill to believe that carbon dioxide is an element?I think not.It is more natural to assume that he has a very shallow understanding of at least one of these terms,and is dependent on the competence of others for determining their reference;the principle of charity seems to dictate an interpretation like that.If Bill understands the lexical meanings of“carbon dioxide”and“element”,he will know that carbon dioxide is not an element,and insofar as he is rational(in addition to being serious and reflected)he will not assent to“carbon dioxide is an element”.So when Bill even so gives his assent,that must be because ofignorance,and then we have an example of a normal English speaker who sincerely, and on reflection,assents to a sentence“p”without believing that p.What Bill does believe,is just that what the experts call“carbon dioxide”is a specimen of what they call“element”.

3.4 Case studies

Letusnowturntoaclassicsubstitutivityproblem,Frege’sEveningStar-Morning Star puzzle.([6])Solving the Fregean puzzle is key to removing the remaining obstacle to quantification into attitude contexts.The Fregean names“the Evening Star”and“the Morning Star”look much like definite descriptions,and that is perhaps the reason why some substitute for them the Latin version of the corresponding Ancient Greek designators(both of which end in“-os”),“Hesperus”and“Phosphorus”,to make them more name-like.In our time,“Hesperus”(or“the Evening Star”)and“Phosphorus”(or“the Morning Star”)are both used to refer to the planet Venus.That these two names are now used to refer to the same object is not something that can be known a priori,and someone who does not know that Hesperus=Phosphorus may believe things about Hesperus that he or she does not believe about Phosphorus,for instance that Hesperus is visible only in the evening.But 3000 years ago in Greece, the linguistic situation was different.At that time,Greek astronomers still believed that the brightest heavenly body(apart from the moon)seen in the evening sky is different from the brightest heavenly body(apart from the moon)seen in the morning sky,and with the names“Hesperus”and“Phosphorus”they intended to refer to two different objects.

In both of the two cases I am now going to analyze,someone believes that Hesperus,but not Phosphorus,is visible only in the evening.But in case number 1,Mary, a person living in our times,believes that,whereas in case number 2,the person who believes it is Agamemnon,who lived 3000 years ago.

Maryisapersonwhoknowsverylittleaboutastronomyingeneralandtheplanet system Earth is part of in particular.She has picked up the names“Hesperus”and“Phosphorus”somewhere,but she does not know that both names refer to the same planet Venus.Instead she believes that Hesperus and Phosphorus are two different stars.But her references to Hesperus and Phosphorus are not direct.Like John, who uses the name“Tully”as short for“the person I have heard being referred to as‘Tully’”,Mary is not able to use“Hesperus”and“Phosphorus”as directly referentialterms,butcanonlyusetheminthedescriptivesensesof“theheavenlybodyIhave heard being referred to as‘Hesperus’”and“the heavenly body I have heard being referred to as‘Phosphorus’”respectively.Therefore the attitude report“Mary believes that HesperusS,but not PhosphorusS,is visible only in the evening,”is not an example of a context where substitutivity of directly referential terms fails,because the terms“Hesperus”and“Phosphorus”as Mary uses them are not directly referential.On the other hand,the report“Mary believes that HesperusR,but not PhosphorusR,is visible only in the evening,”ascribes to Mary the self-contradictory belief that Venus, but not Venus,is visible only in the evening.John and Mary are both what I shall call (following Searle[23])“parasitic users”of the names we are concerned with.

But Agamemnon was not a parasitic user of the names“Hesperus”and“Phosphorus”,and he knew a lot about the astronomic theory of his day.His problem was not that he did not know a priori that Hesperus is Phosporus,because he knew instead infallibly that Hesperus is not Phosphoros.In accordance with the usage of Greek astronomers of his time,he used the names“Hesperus”and“Phosporus”in a different way from the way we use them today,he did not use them to refer to the planet Venus. Agamemnon uses them instead to refer to two distinct objects,each of them supposed to be a heavenly body,and clearly different because Hesperus is visible when Phosphorus is not.Agamemnon was a fully competent user of both names,and able to point out Hesperus and Phosphorus in the sky when they were visible.But as was later discovered,there is only one heavenly body where Agamemnon thought there weretwo.HesperusandPhosphorusdonotexist.TheobjectthatcausesAgamemnon to have beliefs about Hesperus and Phosphorus is Venus,but Venus is not the intentional object of any of his beliefs.This means that Hesperus and Phosphorus are both purely doxastic objects.Again,the attitude report“Agamemnon believes that HesperusS,but not PhosphorusS,is visible only in the evening,”cannot be used an example of a context where substitutivity fails,in this case because the terms“Hesperus”and“Phosphorus”as Agamemnon uses them are not co-referential,and not meant to be.

AsimilarsolutioncanbegiventoQuine’sproblemaboutRalphwhobelievesthat the man in brown hat,but not Ortcutt,is a spy,on the assumption that Ralph uses both terms,the definite description“the man in brown hat”and the proper name“Ortcutt”, to refer directly,which is a possibility.(In lack of a proper name for something,one can use a description to refer directly to it,and it is not a prerequisite for believing something of an object de re that one has a name for that object.)The object that causes Ralph to believe of the man in brown hat that he is a spy,is Ortcutt,but Ortcutt is not the intentional object of his belief.The intentional object is purely intentional, purely doxastic.

My view is that there is no problem of substitutivity failure in attitude contexts whenthetermsinvolvedreallyaredirectlyreferentialasusedbythesubjectwhoseattitude isreported.Allegedexamplesofbeliefcontextswheresubstitutivityofdirectly referential terms fails,can be of two different kinds:In one type of case the subject uses the two terms involved to refer to two different objects,at least one of them purely doxastic and therefore different also from the one object the reporter refers to with them.(If,on the other hand,the subject refers to two different existing objects, we would simply say that the usage of the subject is deviant,and nobody would see it as an example of substitutivity failure.)In another type of case,the subject refers to the same object with both terms,but not with both directly:at least one of the twoterms is used descriptively,as is shown by the examples of John and Mary above.

Kripke’s puzzles about belief can also be easily solved now.First there is Pierre, who assents to“London is beautiful”when this sentence is spoken in French,with the French name“Londres”in the place of“London”,but not when it is spoken in English.Here semantic reference and speaker’s reference have come apart,maybe the referents too.Pierre has learned the English name of London as part of learning to know the city itself,or at least a part of it,through his own experience.He has earlier learned the French name from what his parents told him about the city when he was a child.When Pierre speaks English,or assents to sentences spoken in English, we can take his reference to London as coinciding with the semantic reference(and not only referent)of“London”,and be direct.When he speaks French,on the other hand,he most probably uses“Londres”as short for“the city my parents called‘Londres’”,but there is also the possibility that he refers to a purely doxastic object.Peter, who assents to“Paderewski had musical talent”when Paderewski is talked about as a pianist,but not when he is talked about as a politician,uses the name“Paderewski”parasitically in two different ways since he believes there are two different men who share the same name:in one context as short for“the man people in[group A]refer to as‘Paderewski’”,in another context for“the man people in[group B]refer to as‘Paderewski’”.

3.5 How a proper name can be used as a description

BeforeIconcludethisdiscussion,Ishallsaymoreaboutparasiticusageandother ways of using proper names,which are supposed to refer directly,as short for definite descriptionsthatcontainthenamesthemselves,likethedescription“themountainthe local people call‘Afla’”and the description“the person Bob calls‘Sally’”.

The name“Afla”,together with the name“Ateb”,is used by Frege in a letter to Jourdain,probably from 1914([7],pp.126–129).Both names play important parts in an example much like his example about“the Morning Star”and“the Evening Star”, also designed to illustrate the destinction between Sinn(sense)and Bedeutung(reference),the main difference being that“Afla”and“Ateb”are clearly proper names while“the Morning Star”and“the Evening Star”look more like descriptions.A traveler comes to a place from where a snow-capped mountain called“Afla”is seen from the south.He picks up the name which is then put into circulation.Another traveler comes to a place from where a snow-capped mountain called“Ateb”is seen from the south.He picks up this name,and it too gets into circulation.Everybody thinks that AflaandAtebaredifferentmountainsuntilsomeonemakestheunexpecteddiscovery that they are the very same mountain,seen from different sides.Now I am going to change Frege’s example so that one traveler comes to both these places,first to the place where the mountain is seen from the south,and then to the place where it is seen from the north.He then starts to wonder if it is the same mountain he has seen fromtwo sides,so that“Afla”and“Ateb”are really two names of the same object,but he cannot be sure.Because of this,he cannot use these two proper names as directly referential terms when he is thinking to himself.To him,the names“Afla”and“Ateb”get the meanings of the descriptions“the mountain the local people call‘Afla’”and“the mountain the local people call‘Ateb’”respectively.We would perhaps not call him a parasitic user of these names since he is able to identify their referent demonstratively,but like a parasitic user he can only use the names descriptively.

Parasiticuseofpropernamesisverycommon,andweallusenamesparasitically on a regular base.The following is an everyday situation:Alice and Bob are having a chat.Bob mentions someone he calls“Sally”,an alleged relative of his who seems not to be an acquaintance of Alice’s.Having related something Sally did the other day,Bob remarks,“Sally is smart,don’t you think so?”“Yes,”Alice replies,“Sally is inventive,indeed.”

There is no doubt that Alice makes an assertion by using the words“Sally is inventive,”and there is no reason to suspect that she is not sincere.But Alice,being a parasitic user of the name“Sally”in this conversation,is not in a position to believe anything de re about Sally as long as she is not able to identify the object she refers to in any other way than by the description“the person Bob calls‘Sally’”.She cannot know whether the person Bob calls“Sally”is the same as the person somebody else, say Connie,calls“Sally”,and she cannot be sure that the person Bob calls“Sally”is not someone she herself is acquainted with already,perhaps under a different name. She cannot even be sure that there really is a person Bob calls“Sally”;Bob may be telling her lies.What Alice can believe at most is that some relative of B’s whom Bob refers to as“Sally”,and relates this and that about,is inventive.Using proper names and other terms parasitically in uttering a sentence,one can give expression to thoughts one cannot think oneself.One can also do this by assenting to somebody else’s utterance.There is a gap between assertion and belief.

Minding this gap and the corresponding distinction between speaker’s reference and semantic reference solves many substitutivity puzzles.Those that remain are solved by recognizing doxastic objects as objects we can refer to and quantify over also when they are purely doxastic.

4 Conclusion

In this paper I have discussed what is also the topic of my book Reference and Intentionality:the two big obstacles there seem to be to making sense of beliefs that are directly about objects(de re)and quantification into belief contexts,and I have shown a way of dismantling both obstacles.One obstacle is that the object a belief is supposed to be about,may not exist.I have shown that a viable solution to this problem of possible reference failure is to recognize purely purely intentional objects in general and purely doxastic objects in particular.The other obstacle is that co-referential singular terms appear not to be interchangeable salva veritate in belief and other attitude contexts,even when both terms are supposed to be directly referential. My solution to this problem of substitutivity failure is twofold:first,I argue that demonstratives and other indexicals should not be regarded as directly referential in the relevant sense;second,I argue that it is not semantic reference,but speaker’s reference that is decisive for the question what the speaker believes.The semantic reference of a proper name can be regarded as direct,but the speaker’s reference may even so be indirect,and that is what counts.When we look at speaker’s reference,not at semantic reference,we will find that co-referential terms that are used as directly referential will always be interchangeable in attitude contexts after all.

[1] O.Asheim,1996,“Creatures of imagination and belief”,Nordic Journal of Philosophical Logic,1(1),61–78.

[2] O.Asheim,2011,“What is said with a sentence σ in a context C?”,in T.Wyller,S.G. Carson,J.Knowles and B.K.Myskja(eds.),Kant:Here,Now and How:Essays in Honour of Truls Wyller,pp.161–178,Paderborn:Mentis.

[3] T.Crane,2011,“Existence and quantification reconsidered”,in T.E.Tahko(ed.), Contemporary Aristotelian Metaphysics,pp.44–65,Cambridge:Cambridge University Press.

[4] T.Crane,2012,“What is the problem of non-existence?”,Philosophia,40(3),417–434.

[5] G.Evans,1982,The Varieties of Reference,in J.McDowell(ed.),Oxford:Clarendon Press Oxford.

[6] G.Frege,1892,“ÜberSinnundBedeutung”,ZeitschriftfürPhilosophieundphilosophische Kritik,100,25–50.

[7] G.Frege,1976,Wissenschaftlicher Briefwechsel,in H.Hermes et al.(eds.),Hamburg: Felix Meiner.

[8] R.Ingarden,1960,Das literarische Kunstwerk,Tübingen:Max Niemeyer Verlag.

[9] R.Ingarden,1964/1965,DerStreitumdieExistenzderWelt,Tübingen:MaxNiemeyer Verlag.

[10] S.Kripke,1979,“A puzzle about belief”,in A.Margalit(ed.),Meaning and Use, pp.239–283,Dordrecht:Reidel.

[11] S.Kripke,1980,Naming and Necessity,Cambridge,Mass:Harvard University Press.

[12] S.Kripke,1977,“Speaker’s reference and semantic reference”,Midwest Studies in Philosophy,2(1),255–276.

[13] A.Meinong,1971,“Über Gegenstandstheorie”,in R.Haller(ed.),Abhandlungen zur Erkenntnistheorie und Gegenstandstheorie,pp.481–530,Graz:Akademische Druckund Verlagsanstalt.

[14] M.Michael,2010,“Belief de re,knowing who,and singular thought”,The Journal of Philosophy,107(6),293–310.

[15] F.Moltmann,2011,“On the semantics of existence predicates”,in E.H.Ingo Reich and D.Pauly(eds.),Proceedings of the 15th Sinn und Bedeutung(2010),pp.31–54, Saarbrücken:Saarland University Press.

[16] G.Priest,2005,Towards Non-Being:The Logic and Metaphysics of Intentionality, Oxford:Oxford University Press.

[17] W.V.Quine,1961,“On what there is”,From a Logical Point of View,pp.1–19,New York and Evanston:The MIT Press.

[18] W.V.Quine,1956,“Quantifiers and propositional attitudes”,the Journal of Philosophy,53(5),177–187.

[19] H.Reichenbach,1947,Elements of Symbolic Logic,New York:Macmillan.

[20] R.Routley,1980,Exploring the Meinongian Jungle and Beyond:An Investigation of Noneism and the Theory of Items,Canberra:Australian National University.

[21] B.Russell,1905,“On denoting”,Mind,14,479–493.

[22] B.Russell,1919,“The philosophy of logical atomism”,The Monist,29(3),345–380.

[23] J.R.Searle,1983,Intentionality:An Essay in the Philosophy of Mind,Cambridge: Cambridge University Press.

[24] J.R.Searle,1958,“Proper names”,Mind,67(266),166–173.

[25] P.A.Seuren,2005,“Eubulides as a 20th-century semanticist”,Language Sciences, 27(1),75–95.

指称、意向性和纯意向对象

奥拉夫·阿斯海姆

奥斯陆大学哲学系

南京大学哲学系

olav.asheim@ifikk.uio.no

2016-07-03 Revision

2017-02-24

要使信念和其他命题态度语境的量化有意义，存在两个大的困难。一个是所指称的对象可能不存在；另一个是共指称词项的替换性原则可能会在这样的语境下被破坏。我将在本文中论证，第一个问题可以通过如下方式解决：将信念对象看成是一类实体，这种实体不管它是否存在，都可以被指称。我将概述一种存在物的理论来使其可行。此外，我将论证，指示性指称本身不是直接指称；而且，当所涉及的共指称的两个词项都以直接指称的方式使用时，不存在替换性问题。