Defenders of some kind of description theory about reference, of which I am one, have to deal with Kripke’s rigid identifiers in one way or another. (A rigid designator is something that designates the same individual in all possible worlds.) One way is to deny that names are rigid designators, which can be done by showing that the meaning of a name is something other than its reference. Or we can accept that there are some terms that are rigid designators but explain rigid designation as a special kind of description. I personally think that both are good approaches, meaning that I think names, as we use them, are not rigid designators but simply a description like: “the person who is called X”, where being called X is defined in terms of social conventions, the name the individual responds to, ect, and that I think that if rigid designators are needed to explain certain linguistic phenomena that they can be reduced to some kind of description. Since I have explored the first possibility in other places I will now turn to the second, and in doing so address the criticisms of that approach raised by Scott Soames.
A reason to think that rigid designators are actually rigidified descriptions (a description crafted in such a way that it refers rigidly to an object) is because descriptions are the vehicle by which we know what a term refers too. If I simply gave you a word, terotin, you wouldn’t be able to use that word meaningfully or know what that word referred to, unless I provided you with some kind of description, either explicitly or through context. To say that a word is a rigid designator is to say that we use it to refer to the same individual in all possible worlds. But, if we use it meaningfully to refer to individuals in these other worlds, how do we know whom it refers to in those worlds? Since I don’t believe that language has any special powers that allow it to reach out to the essence of things I think that it must be through some kind of description, simply because of the lack of other options. I propose that the description behind a rigid designator, R, is thus: let P be a description that uniquely picks out x, the object our rigid designator refers to in this world. Our rigidified description then is Ry iff y is a counterpart of the x that is P in W (where W is our world), assuming that an object is considered to be a counterpart of itself in our world. Let us abbreviate this description to be Ry iff y is actually P, where actually means a counterpart of the x that is P in W.
Soames’ first criticism is that rigidified descriptions only pick out the right object when that object exists, and fail to refer to it when it doesn’t exist. He cites the example of referring to past objects as an example of how names, real rigid designators according to Soames, refer to non-existent individuals while rigidified designators cannot. This is only a valid complaint if we hold that time doesn’t exist all at once. Since the most reasonable view of time is that it does exist all at once then rigidified designators can pick out past individuals, since they aren’t constrained to ranging over only the individuals who are currently present. Of course this doesn’t cover the cases of possible worlds in which the individual never existed. But in these worlds it seems equally strange to say that the rigid designator refers to some individual in that world. All we said about rigid designators was that they were to pick out the same individual in all possible worlds, not that they were to pick out that individual even in worlds where that individual doesn’t exist. (And what can it even mean to pick out an individual in a world where that individual doesn’t exist and never has existed?)
His second and third criticisms deal with picking the relevant P, but I will set them aside, since I think that “is called” does nicely.
His fourth criticism is that rigidified descriptions require people in other possible worlds to believe something about our world in order to use rigid designators. Specifically if their world is V then they must believe that y is actually P, meaning that they believe that y is a counterpart of the x that is P in W. But they are in world V, why should they believe anything about W? This criticism turns on the faulty assumption that the propositional content of identical sentences uttered in different possible worlds is the same, which I argued against previously. When someone in V uses a rigid description it is perfectly reasonable to say that what they mean in that y is a counterpart of the x that is P in V. Now if they really are referring to the same individual, and the inter-world counterpart relation is really a kind of identity relation, then these descriptions will pick out the same individuals in all possible worlds. But that doesn’t make them the same description, nor does that imply that the people in V have beliefs about W. And thus no problem arises.
Since Scott Soames is a respectable philosopher I assume that the objections he raises against rigidified descriptions exhaust the reasonable objections to that approach. And thus by meeting them we can conclude that no substantial objections can be raised to rigidified descriptions that can’t be raised to rigid designators in general. For example, the fact that no description of the counterpart function has been provided might be seen as a fault, but the existence of rigid designators require the same function, if there are to be the individuals in different possible worlds who are “the same”, and so it is equally a problem for both. And thus I think that we should prefer rigidified descriptions, if for no other reason then that they give us a way of determining if an individual really is referred to by a given rigid designator or not.