Names -- Not (Just) to Mention Things

In communication (and, I think, in thought), if we want to refer to an object in the world, we can either point to it or describe it. Pointing (ostension) picks out the object directly, and can be done verbally -- even in the absence of the object one could otherwise nonverbally point at -- simply by attaching a semantic marker (a sign or name), by fiat or by convention, to the object in question. Describing works by narrowing the domain of objects that fit the description until one can identify a particular object in a given context, often using a hierarchy of descriptive categories that range from very general to very specific.

We often choose names that are brief descriptions -- e.g., "Ericsson", "waterfall", "fireplace". They allow us to point and describe all at once. Even such brief descriptions, given a narrow enough context, can succeed in picking out an individual object. Other names, even though not chosen for their descriptive value (e.g., "Dweezil", "wall", or "gizmo"), end up serving as descriptions in a rather subtle way: to wit, they describe, self-referentially, how people conventionally denote the objects that the names belong to. Just having the property of being called "Dweezil" by people is enough for one person to be identified to a stranger, who need do no more than observe who answers to the name.

Most of the foregoing account has been anticipated, at least piecemeal, by famous philosophers of language. The notions of "pointing" and "description" correspond to Mill's "denotation" and "connotation", to Frege's "reference" and "sense", and to "extension" and "intension" as used by Putnam and others. The notion of indexicality has been explored by Putnam (environmental indexing) and Burge (social indexing) and others. My view is fundamentally a causal one (the sort explored by Kripke and Putnam), in which a name originates by something like a baptism or dubbing, and subsequent use is intended by speakers to pick out the very same reference (and the same sense, I'll argue). It will be interesting to see, after credit is apportioned, whether any remains for me.... More important, however, is whether the account is supportable by human experience.

The tension between pointing and describing as approaches to reference is reflected in the philosophical debate between direct reference theorists like Mill and descriptive theorists like Frege and Russell. I think the reason the debate could be joined in the first place is that names do not exclusively denote (as Mill would have had it, at least for proper names) or exclusively connote (as Salmon says Russell should have had it [p. 116]¹), but instead manage to denote by means of connoting. This reading conforms to a Fregean view: "[T]o the sign there corresponds a definite sense and to that in turn a definite reference, while to a given reference (an object) there does not belong only a single sign [p. 143]." For Frege, a sense provides a route to the reference; I will argue that only a particular component of the sense does this job, and that other components of the sense are responsible for the many difficulties (or at least many of the difficulties) that philosophers of language have faced with respect to names and the things or kinds to which they refer.

My thesis is that when a name is used to mention something, it does more than just refer to that thing; it also provides the information that the thing referred to has the name used. In other words, a name is implicitly mentioned whenever it is used to mention something. This implicit mentioning confers upon a name some excess semantic baggage -- part of the sense of the name -- that is harmless in ordinary situations, but just weighty enough to cause confusion whenever unusual circumstances cause the reference for the name to be unclear. Two names that refer to the same object, then, differ from one another in that each attributes to that object a different name. And, because they differ in this regard, they may not be interchangeable salva veritate -- unless they are interpreted as shorthand for more complex indexical terms (about which more anon).

I don't think these famous problems of reference would exist if this nondenotative component of names were semantically insignificant. But I believe that the attribution -- self- attribution, to coin a handy term for purposes of this discussion -- of a name to its reference in fact does do significant semantic work. That work is indexical: it provides a social index for the object in question, to the effect that "This is the object I/we/they call "x".² An index, though it serves as a sort of "pointer", does so descriptively -- by citing a relation that holds between the object and some fixed benchmark, such as a time or a place or (in this case) a linguistic community.

A name "x", then, is best understood as shorthand for a more explicit indexical term, "what we call 'x'" or "what they call 'x'" (using "they" in the colloquial way: "They call me 'Mr. Tibbs'."). The "we"³ or "they" indexes "x" to a linguistic community -- either one the speaker readily identifies with, or one s/he otherwise accepts as authoritative in fixing the meaning of the name "x" in ordinary discourse.

Why should this be any better than simply using the name as usual? In this construction, the reference becomes an object indicated by "what". The hidden, implicit self-attribution of a name is replaced by an explicit attribution of that name, within the larger indexical, that is now semantically separate from the reference what, freeing it from the source of confusion.

If "what we call 'x'" seems to apply trivially to any term "x" (and I think it does), I take that as evidence that they are synonymous. And if substituting the more complex term for the name resolves our problems of reference, then I take that as making my case that names have the non-trivial semantic component of self-attribution.

So, let's examine some of the definitive puzzles and thought experiments to see if this sort of substitution really clarifies matters.

Frege's Problem of Informative Identity

Statements of the form "a=b" are not merely trivial tautologies (as "a=a" is), but tell us something beyond the notion that a thing is itself. This comes from a cognitive difference between the two terms of the identity statement, which Frege says stems not from the common reference of the two terms, but from the different senses of the terms.

For Frege, sentences are "names for the True and the False" -- that is, they have truth values. The identity predicate takes two objects as input, yielding a truth value, e.g.:

If we remark upon the fact that "Hesperus" and "Phosphorus" are both names for one planet (Venus), according to Frege, we are dealing with separate senses for "Hesperus" and "Phosphorus". Yet Frege treats these names as having a reference of type "o". Let's see what happens when we treat the names as having references of type "s", knowing that they are intended to refer to something of type "o" -- a perhaps-not-correctly-identified object we can simply call "what".

This maintains (and better, makes explicit) the intuitive notion that two names refer to the same object: the what on each side of the identity statement. The names themselves, then, represent one's sense of an object ... not the object itself.

The Problem of General Substitutivity

According to Frege, when we report another person's remarks, we talk about their sense -- the words have an indirect reference, which is the same as their customary sense. Similarly, when we talk about propositional attitudes (belief, etc.) concerning some object or state of affairs, we refer to a sense. If John (J) thinks that Marilyn Monroe (MM) is lovely, then by Leibniz's Law we should be able to substitute a coreferring term for "MM" -- say, "Norma Jean Baker" ("NJB"). But J might have an entirely different evaluation of NJB's appearance -- so to say "J thinks NJB is lovely" would be to misrepresent J's attitude. If, on the other hand, we make my suggested substitution, we have: "J thinks the person we call 'NJB' is lovely."

This is actually true, if perhaps confusing. For the person we call "NJB" is the person we call "MM", who J truly thinks is lovely. If we also say, truthfully, that J thinks the person we call "NJB" is a homely tomboy, we show that J has contradictory beliefs about the person in question; but this is an understandable state of affairs, given J's ignorance of the MM-NJB identity.

If, however, we use the expression "J thinks MM is not NJB", Leibnizian substitution gives "J thinks MM is not MM" -- which seems to attribute unreasonably a lamentable irrationality to J. Its only apparent interpretation is that J denies the self-identity of a person. But J would certainly not claim that attitude. We know that "MM" and "NJB" refer to the same individual, and we can use the terms interchangeably in statements about our own propositional attitudes. But, within the context of J's propositional attitude, we can fairly substitute only terms that J himself believes to be coreferential.

Making explicit the implicit social index clarifies the subtle difficulty introduced by the inappropriate substitution of publicly (not intentionally for J) coreferring terms: "J thinks the person we call 'MM' is not the person we call 'MM'" is not a statement J would likely assent to; yet it shows that either J could be denying that a person is self-identical (which we can charitably reject) or he could be claiming that two persons are distinct, based upon some confusion over names. Given that we've meddled with the names (without altering either use of "person") in the original statement, this latter option makes a lot of sense.

For Frege, a competent grasp of the senses "MM" and "NJB" would include the fact that they have the same reference. But J has an incompetent grasp of them, and part of each sense is a self-attributed name. Ordinary psychology normally leads us to expect different names to have different references, and the sense J has of each name provides no clue to the fact that just one individual is involved -- so J has justifiably inferred that the names referred to different persons.

According to the prevailing view of causal theory, I take it that J intended but failed to use at least one of the names for its/their customary reference. But something is wrong here. If J's sense of MM picks out any individual, and his sense of NJB picks out another individual, what individuals are they? Perhaps one is "who we call both 'MM' and 'NJB'" -- an actual person in the world (at some time). But we can probably agree that the other is not just some other actual person who J has mistaken for either MM or NJB. J seems to refer to nothing.

Can Frege's view help with this problem? J is referring, on this view, to his senses of NJB (which includes the name "NJB" but not the name "MM") and MM (which includes "MM" but not "NJB"), which senses J would claim have contradictory components -- the most salient of which is that each refers (he thinks) to a different person. Both references are removed from the real world (and are now located in a platonic realm of counterfactuals populated by "might have beens" that differ in various ways from their counterparts in reality). J clearly didn't intend this, and this explanation seems a strange result even for something unintentional (though we'll need something like it later for fictional characters).

I think a variant causal theory works better: J intends to, and actually does, refer in each case to the one individual who we call both "MM" and "NJB". J also intends, but fails, to apply the full public sense(s) to the names "MM" and "NJB". This accounts for the fact that he winds up attributing contradictory properties to one person, seemingly denying her self-identity. This interpretation is quite accessible from my construction, but from the traditional "J thinks that MM is not MM", it seems to be wholly out of reach.

There is also a problem of substitution in modal contexts that I think social indexing helps to dispel. Quine's example (p. 104) is this: "It is mathematically necessary that nine is odd" and "The number of planets is nine" are both true, but "It is mathematically necessary that the number of planets is odd" is false; substitutivity fails to preserve truth value. Now, let me work my magic: "What we call 'the number of planets' is what we call 'nine'." It is now clear that the actual reference for both sides of this identity statement is the number nine; the names "nine" and "the number of planets" are just along for the ride. Substitution merely replaces nine with itself: "It is mathematically necessary that what we call 'the number of planets' [nine, that is] is odd." Note that mathematical necessity doesn't distribute into the "that" clause, giving "It is mathematically necessary that what we necessarily call 'the number of planets' is odd." If it did, substitution would fail for counterfactual situations in which the Sun has an even number of planets. Instead, counterfactuals are ruled out by the construction, which refers to actual things: what we call 'the number of planets' and what we call 'nine'.⁴

True Negative Existentials and Fictitious Names

Consider the claim, "Igor's daughter goes to college," when Igor is actually childless. The reference to a nonexistent person may be puzzling, but the claim doesn't sound merely nonsensical. "Igor's daughter does not go to college" is no better, and, together with the first claim, seems to lead to a violation of the Law of Excluded Middle (p or ~p), since both propositions appear to be false. Russell's solution is to translate these sentences into "Someone is uniquely Igor's daughter and goes to college" and "Someone is uniquely Igor's daughter and does not go to college." He makes explicit the existential claim that there is a daughter of Igor -- which, being false, preserves the Law of Excluded Middle. His intent was to develop a theory of definite descriptions, but I think this amounts to making explicit an implicit index to the real world -- an existential index -- that is embedded in singular terms. Such an existential index is properly part of all names of real objects, but we can usually afford to forget about it in conversation involving only actual things.⁵ Adding this existential index to my construction that already makes explicit a social index, we get "Someone is uniquely the person we call 'Igor's daughter', and (she) goes to college." I won't claim that this contributes much additional clarity to Russell's solution -- but it doesn't hurt....

I think the implicit existential index in names is what makes the problem of true negative existentials seem so sticky. The statement "The present king of France does not exist" would appear to be true only if "the present king of France" refers to nothing at all. If I'm right, and the name includes a Russellian existential index, then we have a built-in contradiction: "Someone (exists who) is uniquely the person we call 'the king of France', and (he) does not exist." Recalling Grice, I think we need to reinterpret what appears to be a structural linguistic break-down, in terms that make sense of our shared commitment to communicate effectively. Happily, I think my social index can help. The problem is that no actual person has been baptized with the name "the present king of France" -- so the name serves as a social index for nothing at all. The fact that names customarily are tags for existent (at some time) objects, for which the implicit existential index is unproblematic, makes the use of a "loose tag" problematic. It is because the presumption that names refer to existent things is built into our language, not because of logical flaws in true negative existential statements, that this problem emerged.

What about fictions? My claim "Figaro is hairy" contains no reference to a real-world character. Instead, it contains a reference to a character from fiction. In such cases, I think Russell's existential index is replaced by a fiction index: "In the stories of Beaumarchais there is uniquely a character called 'Figaro' who is hairy," or "In the story 'Pinocchio' there is uniquely a character called 'Figaro' who is hairy." Whether I mean to describe the barber of Seville or Geppetto's cat depends on which fictional "world" I am thinking about.⁶ Once the fictional context is established with an audience, I can converse within that context with the same success I would experience in a real-world context. Often, a name by itself is sufficient to establish the fictional context (e.g., "Hamlet", "Pinocchio", "Frodo", or "Heathcliffe") -- demonstrating in a special way the semantic power of a name!

Putnam's Twin Earth Thought Experiment

Putnam wants to show that a single intension can indeed have more than one extension associated with it -- something traditionally considered impossible. With his Twin Earth being identical in all respects to Earth except for location and the chemical composition of what the respective denizens call "water", it would seem that all concerned share the same intension while each group has a different extension (H₂O and XYZ, respectively).⁷

Putnam thinks that extension is (usually) fixed socially by a linguistic division of labor. Experts flesh out the social meaning of terms with esoteric knowledge that simply is not possessed by most individuals. Yet ordinary individuals, when they use a given term, intend it to have the full-blown, expert- enhanced social meaning. It strikes me that this has people knowing how they mean, but not what they mean -- an uncomfortable state of affairs, I would think. That's not to say that Putnam is mistaken in this, however; I argued the same thing with respect to J a moment ago. Perhaps this intention reflects a universal commitment to what Grice calls conventional meaning, while downplaying speaker meaning.

Where I take issue with Putnam is this: I think extension is fixed by the social index -- the name -- and not by the fully defined, "expert" intension. This name is part of both the full public intension and a sketchier private intension. We do intend our private intension to match the public one; the fact that we sometimes fail in this means that, while we do refer to the public extension, we contradict aspects of the public intension.

The point of the experiment for Putnam is that extension can't be fixed by intension, and that "'meanings' just ain't in the head! [p. 232]" My interpretation should by now be obvious: The name "water" is used by two different societies as a social index for two kinds of substance; that index fixes a different extension for each community, and is a part of the intension for each community. (So, contrary to Putnam, the extension is fixed by intension -- if only a part of it; indeed, fixing extension is the primary function of a name, and the fact that the extension has the name in question is the part of the name's intension that does the fixing.) The fact that all other aspects of the public intension and the twin- public intension are identical is simply incidental.

Putnam resolves the question of meaning into two problems: determination of extension and individual competence (p. 249). I agree. But I think extension is determined by the name, along with the speaker's intent to be faithful to that name's public extension. And I think individual competence is just a function of how well one manages to match one's personal intensions with their public counterparts -- that is, how well we grasp what Putnam calls stereotypes, which I agree we feel a duty to learn and share.

So, though I agree with Putnam about the importance of indexicality and about the social commitment to shared intension, I am uncomfortable with his conception of the nature of thought (I think somehow it is in the head). But perhaps I'm just being squeamish. If I accept that the extension of a name for a speaker is fixed by his/her intent to match the public extension, why shouldn't I accept that the intension of a name for a speaker is fixed by his/her intent to match the public intension (which intent I have already acknowledged)? My best defense is that I may be more interested in speaker meaning than in conventional meaning, and that only the latter is fixed socially as Putnam claims (more on this issue in a moment).

Burge's Thought Experiment

Burge argues for a different sort of social indexing -- a powerful sort that actually determines the content of an individual's thoughts. Here's a sketch of his thought experiment:

S has arthritis, along with many thoughts about what arthritis is. One day he awakens to a pain in his thigh and thinks his arthritis has spread there. But this can't be, for arthritis is by definition an inflammation of joints, not of tissues.

Now consider Twin-S, who is identical to S in all respects (physically, mentally, and biographically -- their lives are exactly parallel in all respects), but whose society differs in a single respect from that of S: arthritis is defined to include a variety of rheumatoid conditions that include what Twin-S has in his thigh. Twin-S and S have never had any experience that would have informed them whether such a pain in the thigh was or was not what physicians, lexicographers, or informed laymen in their respective societies would call "arthritis".

So Twin-S could not really be thinking that he has arthritis in his thigh. But he would be thinking something true, while S would be thinking something false; so the concepts they each express by "arthritis" are not the same. "Arthritis" in Twin-S's community doesn't mean arthritis, for it isn't restricted to joint inflammation. How, then, could Twin-S have acquired the concept arthritis?

Burge's conclusion is that, despite being identical in all physical, phenomenal, and behavioral respects, S and Twin-S nonetheless have different mental contents -- and the difference is due to the differing social environments.

Let me now bring out the implicit social index in "arthritis". "What I have in my thigh is arthritis" becomes "What I have in my thigh is what they call 'arthritis'." This identity statement will be true if the "what" on each side denotes the same thing. For Twin-S, the extension of "arthritis" is broader than it is for S; they refer to overlapping but different categories by the term. But S and Twin-S both pick out the "proper" extension for their respective linguistic communities: what they call "arthritis". They both attempt to identify this referent with something else -- but what?

The other half of the identity statement, what I have in my thigh, lacks any social index; it is an internal condition brought to awareness by phenomenal experience, and its nature is unspecified -- until what I have in my thigh becomes identified with something more specific. We need expert analysis of this phrase in order to see whether its reference is the same as that fixed by the name "arthritis" in the second phrase. For both S and Twin-S, a trip to the doctor will produce the expert analysis; Doc and Twin-Doc both identify the very same pathology, but Doc notes that it isn't the same thing as what we call "arthritis", while Twin-Doc notes that it is.

Do S and Twin-S really have different thought content, then? I think it all boils down to whether the thing pointed to is part of the act of pointing, or instead is related (by the act of pointing) to the pointer. On the former interpretation, I think Putnam and Burge win the day; by thinking "arthritis", S has, as part of the content of his thought, the concept of arthritis shared by his linguistic community.

I prefer the latter interpretation. Thinking "arthritis" relates S to the extension of "arthritis" via the social index bestowed upon it by his linguistic community. But S doesn't have the concept arthritis. His private intension differs in a small but important way from the public one. Arthritis is what they call "arthritis", not S's conception of what they call "arthritis" -- and it is this latter notion that forms a part of his thought contents. On the other hand, Twin-S has precisely the same thought contents as S, but it so happens that his conception of what they call "arthritis" matches the intension of his linguistic community. Burge claims that Twin-S lacks the concept arthritis, for that is indexed to S's community, not his own. In fact, Twin-S does have one of two concepts for "arthritis" that exist in the universe; Earth and Twin-Earth (to paraphrase Churchill, I think) are two worlds separated by a common name. No linguistic community can lay definitive, universal claim to what "x" denotes ("This is x!"), so long as it is possible for other linguistic communities to baptize other things with the name "x". In fact, it is precisely such universal claims that lie at the heart of paradox after paradox.

Conclusion

I hope these examples have persuasively illustrated the notion of names as social indices, and how making the social index explicit by translating names into a "what they call 'x'" construction can clear up some otherwise difficult linguistic problems. Here is how I think it works:

"What" unprejudically denotes the reference, and "they call 'x'" supplies the social index -- part of the sense or intension of the term. I think it's likely that the "full-blown" intensions societies have for their terms are often accreted around this social index, rather than directly around the reference itself -- particularly in cases when someone first hears a name, and only later manages to identify its reference and gradually flesh out a private intension that might eventually match the public one.⁸

¹All citations are from Basic Topics in the Philosophy of Language, edited by Robert M. Harnish (Prentice Hall, 1994).

²I use the term "social index" even though I am inclined to think that there is such a thing as a personal indexing that can take place in language -- such as when a person coins a word for a new object. The word is useful to the individual in expediting thought about the object, providing a "shorthand" for the description of the object (I include sensory impressions and their cognitive derivatives as nonverbal descriptive elements) that would seem to be the only other (unwieldy) means for thinking about the object. Until such a word is shared with others, it would seem strange to think that it pegs the object to a social group, rather than to the sole individual who uses the name. But I needn't press this point; by allowing the interpretation offered in class by Kobes, in which the individual is both speaker and audience in his own internal "dialog" of thought, we can speak of the indexing as being social (even if it sounds schizophrenic) -- and, better yet, since a newly coined name is at least potentially available to other people, it can be thought of as temporarily or accidentally private, though public in principle.

³"We" can take the "editorial" meaning "I" in special cases.

⁴Quine noted a problem with another sort of substitution (p. 347):

(1) Giorgione was so-called because of his size.

After substitution, (1) becomes

(2) Giorgione was called 'Giorgione' because of his size.

Substituting "Barbarelli" for "Giorgione" in (1) gives us a false statement, but substituting in the first position of (2) works transparently. The difference is introduced by substituting "was called 'Giorgione'" for "was so-called".

Note that (2) has some similarity to my construction for explicit social indexing. Let's carry it all the way through:

(3) The person we call 'Giorgione' was so-called because of his size.

(4) The person we call 'Giorgione' was called 'Giorgione' because of his size.

Again, substitution works in (4), but not in (3). I think that in my construction it is clearer that the "so" in "so- called" refers to the preceding name, not to its reference (the person). Substituting another name for the original completely destroys that reference, and implies a new reference that was never intended.

⁵I should mention that there must also be implicit temporal indices. Our earlier example, "MM is lovely", is hardly a statement anyone would literally assent to upon careful reflection; the image is just too gruesome. The name "Marilyn Monroe", like other historical names, is used with an implicit temporal index to some relevant part of famous lives. We can often speak of MM in the present tense, as though she were still twentysomething and gorgeous, without introducing any practical linguistic confusion.

There's another twist on this temporal index. Perhaps Bob shares the same beliefs J does about MM and NJB, except Bob believes they are indeed one and the same person. Still, he thinks MM is lovely and NJB is homely, because each has a different temporal index: NJB is the pre-professional version of the professional MM, who metamorphosed from the former via some developmental and cosmetic changes. Similarly, Bob might think that Cassius Clay is brasher than Muhammad Ali.

⁶Fictional worlds can be platonic but realistic variants on the real world, quite unearthly alien worlds, or cartoonish worlds with bizarre "natural" laws. All exist only in the minds of people, and are shared by our various modes of communication. I think it makes the most sense to consider the artifacts (books, films, etc.) related to fictions to be representations of their respective fictional worlds, and not the worlds themselves.

⁷Putnam depends upon Twin Earth being metaphysically possible, and I have doubts that it is. The shorthand objection is that I don't think it's metaphysically possible to violate natural laws, and that natural laws depend upon the precise values of natural constants (e.g., i, pi, e, h) which I believe could not be altered even in the minutest way without resulting in large-scale differences in the composition of matter and energy at all levels of complexity. In other words, if things were at all different, they'd be very different, ruling out any possibility of Twin Earth (or Earth, even remotely as we know it, for that matter).

I prefer a more familiar scenario based on elms and beeches, for which Putnam says he has the same intension. Surely there could be two earthly island cultures, one familiar with elms and unfamiliar with beeches, the other familiar with beeches and unfamiliar with elms, who might use the term "elm" for elms and beeches, respectively. If their familiarity with the trees is suitably superficial (not scientific), the intension of the term "elm" could be the same for both groups, while there would clearly be different extensions.

⁸I think I'm striking a compromise position between the Augustinian picture of language learning mentioned in class (in which thoughts are prior to and independent of language, linguistic expressions conventionally match up to antecedently apprehended propositions, and linguistic competence is a matter of correctly making these matches) and the anti-individualist view shared by Putnam and Burge. I think the Augustinian picture best represents early stages of linguistic development; later, new terms that are introduced, perhaps in isolation from any ostensive reference or description, must be matched to their reference and public intensions through social interaction.