If proper names are to be thought of as rigid designators we can’t explain how our use of a name comes to refer to a person by appeal to definitions alone (of the kind: “A” is the person who ….), as that would make proper names simply abbreviations of definite descriptions. We do need some account though; if rigid designators are to be anything more than an intellectual curiosity it must be shown that our use of proper names comes about in such a way that we can legitimately say that the name refers rigidly to some person. To fill this gap Kripke introduced the causal theory of names. In its original version Kripke proposed that a name became “attached” to some object through an initial naming event, and that if we come to know that name through a causal sequence of the right kind then when we use the name we are using it as a rigid designator for the thing that was initially named.
Let us put aside the problematic nature of the right kind of causal chain.* Even if we assume that such chains exist for proper names there are still problems with this theory. Let us pretend that the baby who was named Napoleon was replaced with another baby, who had been named something else, who then went on to do all the things that we think that Napoleon did. It would seem that when we speak about Napoleon we are really speaking about the imposter, and that when we use the name “Napoleon” we are making reference to the imposter. But this contradicts the causal theory of names because according to Kripke when we speak of Napoleon we are making reference to the baby who was named Napoleon, and not the imposter. Although it might be possible for some to accept this consequence it certainly seems unintuitive. And more importantly it means that we can have perfectly understandable conversations about the man who we mistakenly think is Napoleon, communicate information about the imposter, ect, all while referencing someone completely different. And if we accept this we are accepting that the reference of rigid designators makes no difference. And if that is the case there is no point in discussing them, or appealing to them, since they don’t reflect actual the actual use of proper names, exactly the problem Kripke was trying to avoid.
To solve this problem Garth Evans has come up with a second version of the causal theory of names. He proposes that the person who a rigid designator refers to is the source of the majority of our information that we have associated with the name. Thus “Napoleon” refers to the person who is the source of most of the information we associate with the name Napoleon. This causal theory of names thus resolves the problem of imposters, since if the imposter replaced the baby named Napoleon at birth then when we use the name Napoleon we are referring to the imposter, since it is he who is the source of most of the information we associated with Napoleon.
But let us pretend that I am talking to you, and that I have told you that the baby named Napoleon was replaced shortly after birth. If I then tell you that Napoleon went on to become a flower salesman you know exactly whom I am referring to, the man who was originally named Napoleon and then replaced. But how is this possible? If Evans’ theory is correct then whenever I use the name Napoleon as a rigid designator it refers to the imposter. But here I am clearly using it to refer to someone else. Now some might contend that my use of it in that context was not as a rigid designator, but it is hard to see why not. We can’t appeal to the causal theory itself; that would be begging the question. And, aside from the causal theory, it seems just like every other use of name; if we can legitimately claim that it isn’t a rigid designator than we can legitimately claim that names are never rigid designators.
Obviously then context at least partially determines whom we are referring to with the use of a name. We might then to be tempted to attempt to construct some kind of hybrid theory, in which we combine context with a causal chain, but I don’t think this can be done without making names just a kind of definite description. This is because context can be best captured as some set of properties that hold for the person being referred to (i.e. the person originally named Napoleon, the person that appears to be drinking a martini to my friend, ect). And if these properties are involved with reference then what we have are definite descriptions, not rigid designators.
And so I must conclude that proper names are not rigid designators, and thus that rigid designators, if they do exist, are not part of language.
Side note: Another example that shows how context plays in important role in whom we are referring to when we use a name is when we know two people with the same name. In such a case we are able to refer successfully to both of them, usually without confusion, a fact that cannot be explained by naming events, or information about them, alone.
* It is extremely hard to describe what constitutes a valid link in such a chain. We might initially think that it has something to do with a speaker who uses the name as a rigid designator passing it on to another person in the right way. But this seems to be a flawed definition since the name might be passed on in an indirect fashion, though a book, or electronic medium, in code, ect. What if in transmission of the name there is an error, changing one letter to another, but then later there is a second error that changes the letter back to the way it originally was. Is this an invalid causal chain (it certainly seems it should be, since the name communicated isn’t completely dependant on the original name, but instead depends also on the nature of the error)? But if it isn’t valid why does a transmission of the name containing a reversed error and a transmission with no errors have the same end result (results in the same behavior, same use, ect)? Doesn’t this show that our use of the name is the same even if it isn’t a rigid designator, which is the exact opposite of what the causal theory of names was supposed to show?