Kripke's Causal Historical Chain Account of Referring

As I understand him, Kripke is attempting to specify the conditions under which an expression N is a name of an item X for a person S. If one does this, then one can be said to have determined the reference of N for S, that is, to determine the reference of N is to explain the conditions under which N is a name of a particular item. This is not the same as fixing the reference of N which is a matter of initially associating an expression with an item. The latter can be done either demonstratively or by description, but in neither case does the demonstrative or the description provide a synonym of the name or even necessary and sufficient conditions for the name to refer to what it does. For Kripke, names are rigid designators of what they name, but the descriptions and demonstratives typically are not. For example, given what we mean by the expression 'one meter', it picks out the same length in any counterfactual situation we use it to describe, but given what we mean by 'the length of stick S'--where S is the standard meter in Paris--it is not the case that this description picks out the same length in every counterfactual situation. In those counterfactual situations, what 'one meter' picks out need not satisfy the description 'the length of stick S' and, therefore, satisfying that description is neither necessary nor sufficient for 'one meter' to name what it does. The same holds for '9' and 'the number of planets.'

Now Kripke concludes from his observation about the semantics of names within modal contexts that "satisfying certain descriptions" is not how the reference of names is determined. Nor does he accept Russell's acquaintance doctrine, that is, that N names X for S only if S is acquainted with X. Kripke takes the commonsensical view that we have names for things we have never been acquainted with, e.g., 'Genghis Khan', 'Napoleon', 'Avicenna'. So, what alternative account of naming does he propose? Briefly, he adopts what is sometimes called a "causal theory of reference" according to which one uses N as a name for X just in case that use is appropriately linked to an initial "dubbing" of X. There are two central ingredients in his version of such a theory:

(1) an initial baptism (viz., someone's picking out X and stipulating that N is a name of X);
(2) a chain of communication initiated by the baptizer's picking out, stipulating, and communicating, and each link of which involves speakers communicating to others about X by using N and the hearers acquiring the intention to use N with the same reference as those speakers. (Such a chain is said to be reference-preserving.)

Then, a rough approximation of necessary and sufficient conditions for such a theory is the following:

N is a name of X for S iff there is a reference-preserving chain initiated by a baptism of X with N and S's acquisition of an intention to use N as he does is a member of that chain.

Note, that S might be the baptizer and his/her acquisition of the intention is part of what begins the chain. But if the user is not a baptizer, then the user successfully refers only because of one's relations to other users. One's reference is then derivative or "borrowed," and very many names in our vocabulary are like that. In such a case, reference is a social act.

It has been objected that a theory like this fails to give a sufficient condition for reference. Gareth Evans (in “The Causal Theory of Names” (1973)) gives the case of S's sitting in a pub and hearing a number of people speaking of someone named 'Louis'. They happen to be speaking of a French King, Louis XIII, but S does not catch this, in fact, he is so confused that he thinks they are talking about a contemporary athelete. Evans thinks it "outrageous" to say that S is referring to their Louis, Louis XIII, when he utters a sentence like "Louis must be a basketball player," even though S has an intention to refer to what they are referring to was a result of their referential usage of 'Louis'. Regardless of who they are speaking of, poor S is wildly off the mark; misinformed about whom they are talking about, he ends up not talking about Louis XIII at all. He is not successfully referring. So, "intending to refer to what the speaker refers to" is not enough to secure one's own reference to the very item the speaker is referring to.

Take another case. My daughter has two friends named 'Erin' who are almost identical in intelligence, tastes, and aspirations, and she often speaks of them without telling me which Erin she has in mind. I am not always able to guess which one she is speaking of, yet I intend to use 'Erin' to refer to whichever girl she is referring to, and go along with the flow of the conversation by asking questions like "Was Erin happy with her grade?" Am I referring with my use of 'Erin'? My daughter might think I am, and her perception of my 'Erin' tokens might cause her to think of one particular Erin. But I am unable to distinguish that Erin from the other despite my use of 'Erin'. One might think I am picking out a particular Erin through the sense specified by 'the Erin my daughter is now speaking of'. However, sometimes she talks about both Erins at once: "When Erin received her grade she was upset, but then Erin suddenly reminded us that her rival John received a lower grade." I am now lost; which Erin did the reminding? John is the rival of which Erin? Yet I say: "How timely of Erin to think of that!" I am fully aware that either of the two Erins might satisfy the description 'the Erin my daughter is now speaking of' or 'the Erin that did the reminding' or 'the Erin whose rival is John'. As such, I find it somewhat strained to say I referred with my use of 'Erin' given that I am not sure of whom I am speaking. To put it another way, because of my confusion it is not obvious to me that a particular individual is the semantic content of my 'Erin' token and, if so, then my token is not used referringly or does not express a reference I made. To borrow a term from Donnellan, my use of 'Erin' is very much like an attributive use of the name, that is, I am thinking only whichever Erin you are speaking of, did the reminding, is the rival of John, etc., without picking out any particular Erin.

One might put the matter this way. Unlike Frege, Russell had the intuition that merely possessing a definite description that happens to be satisfied by some entity X is not enough to be referring to X or talking about X. Evans shares this intuition when he denies that the Description Theorist's conditions are sufficient for reference. For this reason, Russell did not think that definite descriptions were names despite 'denoting' particular entities, that is, being satisfied by those entities (even though Russell may have been wrong in thinking this to be so of every use of definite descriptions, as Donnellan argued).

Now Kripke explcitly denies that having a description that is uniquely satisfied by X is necessary for one to refer to X with N (the 'Feynman' case), and he also denies that having a description that is uniquely satisfied by X is sufficient for referring to X with N (the 'Schmidt'/'Godel' case). But he does seem to think that intending to refer with N to whatever satisfies a certain description when such an intention is a member of a suitable causal chain, is sufficient for reference to X with N. The 'Louis' and 'Erin' cases challenge precisely that claim. However, this does not challenge the idea that some sort of causal connection between an item X and a use of an expression N must be a necessary condition for reference to X by means of N. Indeed, Evans goes on to offer a rival causal account that accommodates some such condition, but that's another theory.