ABSTRACTS |
|
According to the currently popular theory of reference, a token occurrence of a proper name refers to whatever is appropriately determined in the original name-introducing event which lies at the end of a certain chain of reference transmission the token occurrence extends. The original determination relation is not in general the reference relation. It is at best the naming relation. But the notion of naming remains obscure. The notion of a chain of reference transmission is also inadequate for explaining a certain aspect of our normal use of names. To avoid these and other difficulties besetting the popular theory, a truly direct theory of reference is proposed. It reduces reference to assignment. An assignment is a relation between an object and a token occurrence of a name, is mediated by absolutely nothing, and is effected by a person. Any token occurrence of a sentence of the form 'Jane is F' has a definite truth condition only relative to an assignment; relative to an assignment of object 0 to the token occurrence of 'Jane', it is true it and only if 0 is F.
California State University, Northridge
Suppose Smith is running for public office, and in a discussion about her opponent, Jones, she says,
The cheating liar ought to be in jail.
Using the definite description referentially, she would, according to Donellan, be saying of Jones that he ought to be in jail. But in referring to him as she did, she would also thereby be attributing to him the property of being a cheating liar. This is what I call a referential attribution -- where a speaker uses a description referentially in order to attribute to the intended referent properties indicated in the description. A closer look at the status and mechanics of referential attribution yields a variety of results, regarding saying and implying, referring and attributing, presupposition, implicature, natural and non-natural meaning, metaphor, the referential/attributive distinction, and the semantics of demonstrative descriptions.
Department of Philosophy, University of Haifa, Haifa 31-905 Israel, j. berg@uvm. haifa.ac.il
Suppose that experiences possess phenomenal properties as well as representational ones, and that phenomenal properties can be physically realized in a manner similar to that in which a functional state or property can be physically realized. Suppose further that factual knowledge of any such phenomenal property S (say, knowledge that some experience type e has S) is in principle available event to those in whose experience S does not feature; such knowledge is available to anyone who has to hand a reference-
ABSTRACTS* REFERENCE II/ KARLOVY VARY '94 |
72 |
fixing description of S either in terms of its characteristic causes and effects or in terms of a physical state uniquely realizing it.
Someone in whose experience S does feature, however, ought to be able to express the fact that his experience of e possesses property S in another way, a way that reflects its first-person mode of presentation. He ought to have available a way of referring to S which is not available to one to whom S is not so presented. And it seems that he does. It seems, at least, that he does when S is instantiated in his present experience, for he is then in a position to say, e.g., 'Hearing a seventh in the key of F has this [introspectively identified] phenomenal property'.
Another way of referring to phenomenal properties presented in one's own experience may be through the referential use of definite descriptions: one may refer to some phenomenal property as that nonrepresentational property of an experience which is held constant in that experience across all variations in its representational properties, or conversely, as that nonrepresentational property which varies over time in an experience whose representational properties remain stable.
Both of these ways trade on a assumption that subjects are able to identify phenomenal properties as objects of introspective conscious awareness: phenomenal properties are assumed to be "potentially the objects of immediate, i.e. uninferred, knowledge" (Shoemaker, 1991). But is the assumption true? It has been argued that our ability to identify phenomenal similarities and differences across experiences (an ability manifested in the referential use of such definite descriptions as above) in an ability that we could not possess unless the assumption were true. This is perhaps unpromising as a defense of uninferred knowledge of phenomenal properties. In any case it rests on a false general claim about assertions of similarity: in at least some cases assertions of similarity are primitive and unsupported by independent identifications of shared properties. It may be true that phenomenal properties are accessible as objects of introspective awareness, but that truth is no necessary condition of our ability to observe phenomenal similarities and differences.
St. John College, Oxford, OX1 3JP, UK
I will try to motivate the view that we don't need anything more than a disquotational notion of reference, and therefore that the whole idea of a theory of reference is a mistake. Then I will address a major worry about this viewpoint: that much work in the theory of reference has proved quite illuminating, and it is hard to see how this can be so if the idea of a theory of reference is misconceived. I will argue that wile much work in the theory of reference (for instance, that of Kripke and Putnam) does indeed make an important contribution, still the real value of this work is best seen if we view is as not primarily about reference at all but about two other topics; and that the appearance of being about reference can then be explained by the disquotational properties of reference. I will also discuss David Lewis's paper on theoretical terms, both from the point of view of the theory of reference as standardly conceived and from the point of view of the "inversion" of it that I am proposing.
Oxford, Great Britain
previous page
ABSTRACTS* REFERENCE II/ KARLOVY VARY '94 |
73 |
According to Jerry Fodor's Theory of Content (MIT Press 1990), a token of the mental symbol "'cow' means cow if (i) there is a nomic relation between the property of being a cow and the property of being a cause of 'cow' tokens; and (ii) if there are nomic relations between other properties and the property of being a cause of 'cow' tokens, then the latter nomic relations depend asymmetrically upon the former." (p. 93)
I would like to defend the following claims concerning this theory.
1. I propose to interpret Fodor's theory in this way: its core is the claim that meaning is conferred upon mentally represented concepts by semantical laws (SLs) which are a special kind of laws of nature. The "asymmetry condition", central to Fodor's solution of the "disjunction problem", can be seen as a mere consequence of the existence of SLs. The asymmetry condition is equivalent to the statement that exceptions to a SL, as well as instantiations of more complex laws containing the SL as a conjunct, are ontologically dependent on the SL, but that the regular instantiations of a SL depend only on the SL itself. (Being a premilimary, this point is not argued for at length.)
2. A theory of mental content which is based on the postulate of SLs is superior to a theory which invokes only actual causal relations, but not possible ones.
3. According to a nomological theory of content (in its form proposed by Fodor) the denotation of a "natural kind" in a scientific sense. If turns out that if such a concept is acquired in extra-scientific circumstances, its extension is a class of objects (in the case of "count-concepts", like cow) or of stuff (in the case of "mass concepts", like gold) which have a phenomenological property in common, but not necessarily a scientific one. Yet, this doesn't show the concept to be disjunctive, as Fodor claims, except in virtue of the question-begging stipulation that the extension of such a term must be determined in terms of scientific predicates.
4. Even if we take a "natural kind" concept as a scientific concept whose extension is determined by the natural kind some stuff or objects actually belong to, it is in general still indeterminate. The same stuff or object belongs to different natural kinds, depending on which particular scientific taxonomy is considered relevant. We need to distinguish the (phenomenologically determined) extension of water as a common sense concept, from at least two different scientific extensions: "heavy water", i.e. water molecules containing the hydrogen isotopes D and T, belongs to the extension of the chemical concept water, but not to that of nuclear physics.
5. The extension of a concept as it is represented in a cognitive system, is contingent on the categorizing mechanisms of that system. I argue that Fodor's conviction that a nomological theory of content is necessarily verificationist, is due to a misunderstanding of the implications of this contingency. It is a matter of empirical research to find out about the extension of a given (represented) concept, but that doesn't make its content itself depend on the finding out. This becomes clear if one takes care to distinguish the representational system under examination from the researcher's trying to learn about the extension of a
ABSTRACTS* REFERENCE II/ KARLOVY VARY '94 |
74 |
concept entertained by the former. The extension of the former's concept does not depend on the latter's finding out about it, yet the theory were verificationist only if it predicted that this were the case.
6. I have tried to show that the nomological theory presented by Fodor can't predict the extension of a natural kind concept represented by a particular cognitive system, in a non-ambiguous way. The precise determination of the extension of a concept is fixed by the particular categorizing process linked to it. A categorizing mechanism is essentially a triggering device which is sensitive to a specific property, or combination of properties.
CREA, 1 Rue Descartes, F -- 75005 Paris, France
In this paper, I am concerned with a set of issues that revolve around that of direct reference. In the first section, I set out reasons for thinking that we cannot avoid commitment to the existence of direct reference in thought. The argument rests on two assumptions. The first is that we often know that we have referred to spatio-temporal particulars. The second is that large scale spatio-temporal symmetries are at least epistemically possible. In the second section, I consider what minimal set of capacities for direct reference would suffice for the referential abilities which we know we have. I argue that we need six distinct capacities for direct reference in order to be able to directly refer to spatio-temporal particulars other than ourselves. These are the capacity for direct reference to the self, to a time, to a temporal direction, and to three distinct spatial directions. In the third section, I argue that our capacities for direct reference are restricted to those identified in section two, and all reference besides the sorts there identified is reference by description. We refer to an object by description when we refer to it as the denotation of a Russellian definite description. In the fourth section, I consider what kind of epistemic capacities are required in order to understand our ability to refer to spatio-temporal particulars, in the light of the preceding, taking as my starting point for investigation Russell's analysis of knowledge by acquaintance. In the fifth section, I consider the implications of my analysis of direct reference in thought for direct reference in language, and the semantics of propositional attitude attributions. In the final section, I consider some implications of my arguments for issues in the philosophy of mind, such as reductionism, and the causal relevance of thought content to our behavior.
Florida, USA
ABSTRACTS* REFERENCE II/ KARLOVY VARY '94 |
75 |
According to the recent debate concerning the principles of reference determination in the case of demonstrative utterances there are three relevant theories: the intentional, the causal and the contextual theory of reference determination. Since the three theoretical frames are problematic, I suggest an alternative, the interpretational theory of reference determination.
The basis for my proposal is a list of detemonstrative utterances, each marked with the referent of the word 'that' according to the intuitions in our speech community. These examples count as criteria of adequacy that a theory of reference determination should fit. They allow me to show the weakness of the three views. I discuss the intentional view suggested by Kaplan and the one suggested by Marga Reimer (the quasi-intentional view). Furthermore, I criticize the contextual view proposed by Wettstein. Wettstein's proposal is the basis for some modification that lead to the interpretational theory is that one should consider a variant of the principle of charity (or a principle of reasonable interpretation) as the main principle of reference determination. To make this view viable, I have to introduce a normative interpreter, on the one hand, and to explain how one can account for the relevance of special mutual background knowledge between the speaker of an utterance and the actual addressee, on the other hand.
Universität Bielefeld, Deutschland
One expects, and one should, that the principles of logic hold no matter what. The laws of logic should hold for any circumstances, conditions, or situations, and the rules of logic should be truth preserving no matter what one's premises assert. One of the more extreme "no matter what" cases concerns the situation in which nothing exists. That the principles of logic hold for the case of the empty domain is the concern of Inclusive Logic. Recall Russell's remark about testing logical theories with puzzle cases.
In this paper a puzzle is stated concerning the empty domain and arguments are presented with conclusions bearing on the logical form of existence claims. In particular, the standard connection between existence sentences and quantification (Frege, Russell, Quine, etc.) is called into question and an alternative proposal for the logical form of existence sentences is offered to replace the quantificational one.
Queens College and the Graduate Center City University of New York
ABSTRACTS* REFERENCE II/ KARLOVY VARY '94 |
76 |
In Partee (1989) it was argued that the vocabulary of natural languages cannot be
neatly divided into terms corresponding to constants, variables, and deictis, but that
many words have meanings that combine these properties, as illustrated by the following
examples:
(1) (a) Every man faced an enemy.
(b) Most Europeans speak a foreign language.
The word "enemy" in (1a) and the word "foreign" in (1b) are both context-dependent when used as one-place predicates; their interpretations may anchor onto the discourse context (e.g. enemy of "ours", foreign relative to the speaker), or they may anchor onto the "bound variable context" established by the quantifiers in (1a-b): an enemy of the (variable) man, foreign to that (variable) European. (Note the multiple possible interpretations of "Most foreigners speak a foreign language.")
Whether a given context-dependent element can be interpreted as dependent on a given quantifier depends on the linguistic structure, in ways that are very similar to the constraints on the patterning of possible interpretations of anaphoric elements such as the pronoun "he", which may have deictic, "coreferential", "donkey-pronoun", and "bound variable" uses.
This talk will address prospects for the integration of variable-binding and context-dependence, taking advantage of recent work in dynamic semantics, treating meaning as "context-change potential". The relation of "context-change potential" to David Kaplan's "content" vs. "character" distinction will be broached.
The talk will conclude with some remarks on the significance of topic-focus articulation in both pragmatic and dynamic semantic interpretation.
University of Massachusetts, Amherst
The first part of this paper extends a theory of meaning for sentential clauses to a theory of saying and asking for declarative, imperative, and interrogative sentences. Examples of the results are:
'Snow is white' says that snow is white.
'Close the door!' says (for you) to close the door.
'Did Bill walk or did Mary give him a ride ?' asks whether Bill walked or Mary gave
him a ride.
In the second part it is maintained that there is an ambiguity in sentences such as 'Treat each customer politely'. On one reading a use of his this sentence results in a single universally quantified command. On another reading a single use of the sentence results in multiple commands, one per customer. It is maintained that this is not a haphazard pragmatic issue; the readings are conventional, and are provided by our grammar. In the third part of the paper this ambiguity is extended to imperatives and to declaratives.
ABSTRACTS* REFERENCE II/ KARLOVY VARY '94 |
77 |
It is further generalized to uses of NP's other than universal ones. If the NP is of the form 'the N', then the resultant ambiguity coincides with the referential/attributive distinction for definite descriptions. Questions are raised about extending the ambiguity to other NP's (such as indefinites) and to other scope-bearing grammatical items, resulting in a theory of conditional assertions, commands, and questions, and distinctions between assertion and denial, and between requiring and prohibiting.
Irvine
Only a small number of the meaningful sentences of a natural language ever gets tokened. It is a challenge to any account of natural language to explain what fixes the meanings of all the untokened sentences. This is what David Lewis has called the problem of meaning without use. It is clear that untokened sentences cannot get their meaning by their use, nor by the intentions agents have towards those sentences or their utterers. This problem has been discussed in relation to the more general language-relation problem -- the problem of specifying what relation must obtain between a population P and a language L in order for L to be a language of P. The language-relation problem was given definitive form by Lewis, whose solution was shown by Stephen Schiffer to suffer from the problem of meaning without use.
One approach to the problem is to suppose that natural languages have a compositional semantics, i.e. that the meanings of sentences are (somehow) compositionally determined from a finite base of meaningful expressions, all of which occur in tokened sentences. However, Stephen Schiffer has argued that natural languages do not have compositional semantics. Thus, the problem of meaning without use poses an especially serious challenge, if Schiffer is correct.
Schiffer has recently proposed to answer this challenge in a way that will also provide a solution to the language-relation problem. Schiffer's strategy is to reduce the problem for public languages to that of specifying the language relation for languages of thought -- specifying what it is to think in a language -- and then to give an outright characterization of thinking in a language with special provision to avoid the problem of meaning without use.
However, I will show in a precise way that Schiffer has not solved the problem of meaning without use (and so, too, has not solved the language-relation problem). Using Schiffer's characterization of what it is to think in a language. I show that if an agent thinks in some language L, then there is an infinity of languages that the agent also thinks in with the very same sentence tokens, but with arbitrarily different meanings. The sentences on which these languages differ are one and all untokened, and so the difficulty is a version of the problem of meaning without use.
University of Florida
ABSTRACTS* REFERENCE II/ KARLOVY VARY '94 |
78 |
Human beings perceive the outer world through the patterning of their language. In European structural linguistics, this patterning has been understood as the meaning of the sentence, i.e. as a level devoid of synonymy and ambiguity, where, however, the vagueness of meaning is present. Criteria which help establish this structure comprise the substitutability 'salva veritate' of synonyms (except in quotational contexts, which include mentioning of meaning) and the certainty that the speaker is able to distinguish between the meanings of an ambiguous expression.
The shape of the meaning of the sentence differs from the logician's constructions. No such means as overt variables with their explicit scopes, prenex quantifiers, type theory or lambda calculus are present here. The representation of the sentence at this level has the shape of a dependency based network (valency grids with obligatory and optional participants and adjuncts), including further dimensions for coordination (and apposition) and for the topic-focus articulation. This articulation may be viewed upon as the 'deep word order' with a dichotomy of 'given' and 'new' information, expressed by an interplay of surface word order and the intonation pattern of the sentence. The deep word order is relevant for a presupposition being triggered (in topic) or not (in focus), and for the scopes of operators.
This level of sentence meaning may be considered to constitute a suitable starting point for a semantic (-pragmatic) interpretation of the sentence, i.e. of an analysis of its cognitive (ontological) content. Among the main steps of this analysis there is the specification of reference of the referring expressions in the sentence. Various modes of reference have to be taken into account (with definite, indefinite, specifying an g generic noun groups, with verbs and their tenses, aspects and modalities, and so on), as well as the different types of categorization studied by cognitive linguistics). Reference, based on the naive ontology, depends on the structure of the sentence and also on inferencing, using contextual and other knowledge.
Only when reference is specified can we processed from a description of the sentence structure (meaning) to that of the intension (the truth conditions) of the sentence (or of its situational value) and then to the impact of the utterance of a sentence as an operation on the hearer's memory. This impact includes the effect of the speaker's intention, with 'mutual knowledge' and the layers of speech acts and of illocutinon and perlocution. Since inferencing is limited by the hearer's capacities, the report on an attitude can only be fully faithful if it reflects the meaning of the original wording; thus, intersubstitutivity is restricted to synonymy for the 'hyperintensional' contexts.
Prague
Referring expressions like "this house" pick up segments of reality, or parts of experience. Some of these expressions are about concrete things or facts, while others point out to conceptual constructions. Referring expressions may refer determinately or indeterminately. Names and description are determinately referring expressions, whereas "something", "everything", "some", "not all" have indeterminate reference.
ABSTRACTS* REFERENCE II/ KARLOVY VARY '94 |
79 |
Determining the reference may turn not to be problematic also in cases of determinate reference. Following Quine (1979), one may speak of failures of reference; these are possible everywhere.
By "determining the reference" one may understand the following: (i) specifying the referring expression, (ii) correlating it to its referent, i.e. to the object which is referred to by the expression, and (iii) saying how the expression refers to the referent. Kripke (1980) uses the expression "fixing the reference (referent) of a term." He distinguishes between definitions which fix the meaning (by giving synonyms) and those which fix the reference. He manages to show that reference-fixing definitions are more problematic than one might think. Kripke concentrates on problems of correlation within an epistemological framework.
Some expressions refer to reality only through referring to themselves, e.g. "this", "here", "now", "I", "you." In order to determine their reference, one has to know who did utter them and where and when. Russell (1980) calls them 'egocentric particulars.' Reischenbach (1947), who developed Russell's idea, uses the term 'token-reflexive words' of words which refer to the individual signs which express these words. One may wonder in which sense the expression "refer to" can be understood in connection of egocentric particulars or token-reflexive words.
To determine the reference of an indeterminately referring expressions is as difficult
as it sounds. Universal quantifiers are connected to the problem of infinity in
mathematics; also otherwise one may wonder what is 'all' all about. Furthermore, take the
so-called I and 0 expressions from the traditional square of opposition in syllogistics:
while "Some people are literate" is obviously about literate person, is
"Some people are not literate" about literates, illiterates, or both?
References
Kripke, Saul: Naming and Necessity. Harward University Press, Chambridge, Mass. 1980 (1972)
Quine, Willard Van Orman: Word and Object. The M.I.T. Press, Cambridge, Mass. 1979 (1960)
Reischenbach, Hans: Elements of Symbolic Logic. Macmillan, New York 1947
Russell, Bertrand: An Inquiry into Meaning and Truth. Allen & Unwin, London 1980 (1950)
University of Helsinki
For some thinkers the realism/anti-realism debate, in most of its current forms, is misconceived. They say, relying usually on the ideas of Wittgenstein, that there is a way of resolving in which does lead neither to full-fledged realism nor to anti-realism. This line of thought is defended, among others, Jane Heal. She distinguishes between week and pre-philosophical minimal realism and three ways of justifying it: mirroring realism, pragmatism and quietist realism. She finds the first two ways of justification, embodying the common realist and anti-realist strategies, untenable, and opts for quietist realism, according to which we should stop, by becoming aware of the interlocking complexities of our thought and action, our search for an external justification or criticism of minimal realism. The paper suggests that questism is an attractive option only at the first sight. The main weakness of quietist realism lies in the holism which it involves, and tin the further difficulties with its self-reference. It seems that in the end of the day this view must
ABSTRACTS* REFERENCE II/ KARLOVY VARY '94 |
80 |
acknowledge the rightness, at least to some extent, of full-fledged realism or perhaps anti-realism. If not, it will become an unmotivated resignation from taking part in the realism/anti-realism debate and lose all of its attractiveness. The result of these criticism may be seen as a support of the general conclusion that a fully satisfactory and intermediate position between realism and anti-realism is not an available option.
Dept. of Logic and Theory of Knowledge, Catholic University of
Lublin, POLAND
Concerning the semantic of proper names, two main views contrast each other: the indexical view, according to which a name is a context-sensitive expression, and the 'object-dependent' view, according to which diversity of referent is a criterion of difference for names, even though they may sometimes be omophonical.
Recently, F. Recanati has provided an argument for supporting the former view to the effect that the conventions which associate a name and the object(s) for which it stands are social, not linguistic. Indeed, a speaker who does not know a convention of this kind cannot be regarded as linguistically incompetent.
This argument, though powerful, does not seem to automatically apply to either nicknames or idiolectic names, insofar as their use is by definition not grounded in any such convention. A more comprehensive argument is therefore required in order to support the indexical view even when such names are concerned.
Such an argument purports to show that the indexical view is, from a theoretical point of view, more economical than its competitor. The point is that a defender of the 'object-dependent' view has to draw a conceptual distinction that one can dispose of once one passes to the indexical view. The distinction I have in mind is D. Kaplan's distinction between individual and generic names: the former are those that are object-bounded, whereas the latter are, so to speak, what all names which stand with each other in a relation of omophonicity have in common. In actual fact, Kaplan has to draw such a distinction in order to deal with the following datum (firstly discovered by T. Burge): names have predicative uses. This datum indeed turns out to be problematic for the 'object-dependent' view when omophonical names are at stake. For example, David Hume, David Hilbert and David Kaplan are all Davids. Thus, although for the 'object-dependent' view the first "David" is a different name from the second and the third, in that they respectively have different referent, there is something they must share which explains the fact that Hilbert, Hume and Kaplan are all said to be a David. Kaplan accounts for this by holding that these different individual names are all instances of one and the same generic name David.
However, if one maintains that names are indexical in nature, one does not have to draw such a distinction. As L.J. Cohen has shown, a full-fledged indexical may have a predicative use, akin to that of proper names: for instance, the three guys I am talking about are all hes. But this does not compel one to endorse a Kaplan-like distinction between individual and generic indexicals, thereby adopting an 'object-dependent' view of indexicals. As far as an indexical is concerned, Kaplan himself holds that it is a context-sensitive expression: it is not the indexical qua type, but rather its tokens, which changes referential meaning
ABSTRACTS* REFERENCE II/ KARLOVY VARY '94 |
81 |
when used in different contexts. Thus, to account for the predicative use of an indexical, suffice it to say that this indexical expresses in such a suse a property corresponding to the Kaplanian character is possesses qua type.
University of Palermo, Italy
In the present paper I wish to answer the question: what is the reference for a thought? The domain of my considerations will include the mental sphere in which there is a room for thoughts.
For the purpose of my paper I will make two semantically important assumptions:
1. about intentional character of thoughts where the notion "intentional" I take in the sense of Brentano according to whom "intentional" means "directed to something", and
2. about a propositional character of thoughts.
According to denotational semantics I will assume further that propositions refer to states of affairs.
I will make a further step taking into account this part of our mental sphere in which thoughts play the most important role semantically and epistemologically, and pragmatically (in the sense of action). This is the sphere of our beliefs and desires. Beliefs and desires can be treated as our mental states (attitudes) which are realized as particular event tokens by propositional attitudes "Iokne believes that...", "ones desires that...".
The moment of a mental attitude, i.e. C(MA) is a thought the same as the content of a propositional attitude is a proposition. In this outline I want to discuss the idea of the content of an individual mental attitude such as for example somebody's belief.
The position that I will take in the paper can be defined as follows:
As regard treating the content of a mental attitude as a thought I am internalist, but in regard to the problem of reference I stand on a position of objectivism. That means that the reference of a thought is external to it although I do not exclude that a thought can refer to another thought or, more generally, to something that belongs to the sphere of mentality.
I will attempt to show that the problem of reference of thoughts engages us in possible worlds semantics because thoughts as C(MA) refer to states of affairs occurring in possible worlds. I will also try to show that the universe of possible worlds should be extended to fictitious worlds; this holds true in the case of our beliefs that deal with fictious objects. Further I want to prove however that although the tools of possible worlds semantic can be useful in the analysis of thoughts, it is not sufficient to account for the problem of reference of thoughts. Possible worlds analysis does not solve the problem of reference of thoughts whose objects of reference are mental.
Department of Logic and Theory of Knowledge, Catholic University of Lublin, Poland
ABSTRACTS* REFERENCE II/ KARLOVY VARY '94 |
82 |
Looking at the ceiling of the Sixtine Chapel you utter the above sentence. To what do you refer? Do you refer to a spot of paint, saying of it that it is Adam? Or do you refer to Adam, implying that you see him? If Evans and McDowell are right, your sentence means a singular proposition (one that includes its object); but if Adam never existed how can he be a component of a proposition, and if he is not, how does your sentence make sense?
Walton denies that by uttering the above sentence your refer to anything; he also denies that it expresses a proposition. I think that is absurd. First, what you said is true, so you must have expressed a genuine proposition. Second, if what you did was pretence, what you pretended to do could be Swimming just as much as Referring. Third, to pretend to A it must be logically possible to really A, so if it is impossible to refer to Adam you could not pretend to do it, either.
I also reject the view that your sentence is short for "this is an Adam-picture". An object x can be directly seen through binoculars, on television, in a movie, or in a photograph; there is no reason to hold painting deficient in this respect, allowing us to see not x but its representation only. In all these cases we see x by seeing y, and I suggest that this happens when y is regarded as ontologically parasitic on x. Consider the sentence, "That spot over there is my house." Does 'that' refer to the spot or to my house? To both, for my house is that spot. Do I, then, live in a spot? No; the spot is only a way my house looks from here. Yet in a domain D (say, a domain of visual phenomena) the spot is all that the house is: the spot is the-house-in-D.
If Adam is a merely possible being you cannot see him, but you can visualize him. Now if you can see your house by seeing a spot (its D-instance) you can visualize Adam by seeing some paint on the ceiling. Your 'this' refers to both the paint and Adam. How can a bit of paint be Adam? I answer that Adam is a man in the possible world C, but in D (the real world) he is only paint. A man in C may be a spot of paint in D just as a big house in C may be a mere dot in D. Transworld identity is fixed by the fact that Adam-in-D (the painting) is determined by what Adam is in C, as the spot is determined by what my house is in C. Hence seeing x in D is a way of visualizing x in C.
Adam can be a component of singular propositions though he is unreal because he has real instances (pictures, mental images) that can ingress in thoughts. Thus Fodor's constraint, that an explanation of behavior be confined to internal factors, is met: the thought [F(x)] can include an extramental, or even an unreal, entity x, if x has instances that are mental representations or phenomena, and hence can enter thoughts.
The Hebrew University of Jerusalem