Lycan discusses three approaches to proper names.
Russell suggests that proper names should be regarded as abbreviated descriptions, so that his theory of descriptions actually applies to names as well.
Simple version of this view: a name abbreviates a single, determinate description. Which description? The easiest answer is, the description that collects everything I believe about the referent of the name. If I think that Evan has properties F, G, and H, and those are the only properties I think he has, then when I say "Evan is in class today," that means:
There is at least one person who has F, G, and H,
There is at most one person who has F, G, and H,
Everyone who has F, G, and H is in class today.
Problems for the simple description theory:
- It seems to pack too much into the meaning of the sentence. (a) I don't need to have everything I believe about Adam in mind in order to know what I mean when I say that he is in class today. (b) Also seems to get the facts about logical implication wrong. Suppose that property F mentioned above is the property of being a future philosophy graduate student. Then "Adam is in class today" logically implies "There is at least one future philosophy graduate student." But surely it doesn't!
- Different people associate different descriptions with the same names. If the simple description theory were true, therefore, proper names would be ambiguous: they would mean different things in the mouths of different people.
As a result of problems like these, Searle and others proposed a more complex description theory, the Cluster Theory. On this view, a name refers to whatever satisfies a sufficient number of the descriptions associated with it (but not necessarily all of them).
But there are problems for any description theory, including the cluster theory:
- Modal objections: if the description theory is true, then it is a necessary truth that the person one refers to by a certain name satisfies all or most of the descriptions one associates with that name. But in fact most or all of these propositions seem to be contingent. [Lycan's "Objection 3," p. 39.]
- Semantic objection 1: the referent of a name may not satisfy the description I associate with it. (Goedel/Schmidt.) [Lycan's Objection 4, p. 40.]
- "Some people are unaware that Cicero is Tully." The Russellian analysis of the meaning of this sentence, assuming that the person who utters it knows that Cicero = Tully, seems wrong. Such a person associates the same descriptions with both names. ("Some people are unaware that there is exactly one person with F and there is exactly one person with F and whoever has F has F.") [Lycan's Objection 5, p. 41.]
- Semantic objection 2: the description associated with a name may fail to pick out a unique individual (but we can use it to refer anyway). [Lycan's Objection 6, p. 42.]
- fictional names. (All sentences involving them will be false.) [Lycan's Objection 7, p. 42.]
This is basically the idea that names do not have "senses," but refer to objects without the intermediary of a meaning that users need to understand.
- Substitutivity. [Lycan, p. 50.] If the only semantic content of a name is to point to its referent, then every name that refers to the same object has the same semantic content. So coreferential names should be intersubstitutable in any sentence, without changing the truth or falsity of the sentence. But consider belief sentences like this: "Albert believes Mark Twain wrote Huckleberry Finn." It seems that this could be true even though "Albert believes that Sam Clemens wrote Huckleberry Finn" is false. (How does the Direct Reference theorist reply? Main response: bite the bullet and say that despite appearances, names are in fact intersubstitutable.)
- Frege's Puzzle (informative identity sentences) [Lycan, p. 52]
- reference to nonexistents [Lycan, p. 52]
- negative existentials [Lycan, p. 52]
This is the "picture" of the way reference gets fixed that is developed in Kripke's lecture 2.
- empty names [Lycan's Objection 1, p. 56]; reply: in these cases the causal chain just goes back to a failed dubbing; the rest of the chain works as usual
- reference change [Lycan's Objection 2, p. 56]
- misidentification of the object named (the cat named Liz) [Lycan's Objection 3, p. 57]
- impossibility of categorial mistakes [Lycan's Objection 4, p. 58]
We need to distinguish two senses in which we could use the term "meaning": what's in the head of a competent user, and what determines reference. It has traditionally been thought that these are necessarily the same, but on the Kripke/Putnam account, they can come completely apart.
Last update: October 22, 2014
Curtis Brown | Philosophy of Language | Philosophy Department | Trinity University