Causal Descriptivism and Deferential Descriptivism and Meta-Semantic Descriptivism
*) Oskar Minkowski discovered that dogs whose pancreas is removed develop the symptoms of diabetes.
Suppose this is the first time you've heard the name "Oskar Minkowski". Cases like this are good candidates for causal descriptivism. According to causal descriptivism, my utterance of (*) is true iff there is a person standing at the origin of a certain chain of communication leading to my present use of "Oskar Minkowski", and this person discovered that dogs whose pancreas is removed develop the symptoms of diabetes. This comes close to many people's intuitions about possible cases.
By comparison, you probably won't regard my utterance of (*) as true iff the person who first discovered that dogs etc. first discovered that dogs etc.: surely what I said could turn out false. Nor do you regard my utterance as true iff the person who [here follows some rich description of Oskar Minkowski] discovered that dogs etc.: you don't even know any such rich description. Nor, finally, do you have no opinions at all about the truth of my utterance conditional on other propositions.
But these are not the only alternatives to causal descriptivism. Two others are deferential descriptivism and meta-semantic descriptivism.
According to deferential descriptivism, you'll regard my utterance of (*) as true iff the conditions under which I regard the utterance as true are satsfied -- or, more precisely, iff there is a condition Phi such that 1) I regard (*) as true iff the obejct that satisfies Phi discovered that dogs etc., and 2) the object that satisfies Phi discovered that dogs etc.
According to meta-semantic descriptivism, you'll regard my utterance of (*) as true iff there is some person called "Oskar Minkowski" who discovered that dogs etc.
The best account may well be some weighted mix of all possibilities: you regard my utterance as clearly true iff it is true according to causal and deferential and meta-semantic and traditional, qualitative descriptivism; and you'll still regard the utterance as true (albeit no longer clearly true) if one or two of these conjunctions fail, or slightly fail.
Anyway, let's look at how the ingredients should be weighted.
First, causal versus deferential descriptivism. We have to think of cases where (*) would be judged true according to causal descriptivism and false on deferential descriptivism or vice versa, and then see which answer conforms better to our intuitive judgements. That isn't easy because causal descriptivism is notoriously underspecified.
The pet aardvark: Coming across the name "Oskar Minkowski", I decide that this would be a good name for my pet aardvark. Then I become very confused and believe that my pet aardvark once discovered that dogs whose pancreas is removed develop the symptoms of diabetes. To tell you that, I utter (*).
Under these circumstances, is my utterance true or false? This question is less straightforward than it may appear. Of course, the belief I expressed with my utterance is false. But if I am wrong about what the sentence means, I can express a false belief by uttering the sentence even if what I utter is true. I'll return to this possibility below. But for now it doesn't matter because causal descriptivism and deferential descriptivism give the same answer to the present case anyway: that my utterance is false. This is because intending to refer to something else breaks the normal causal chain according to causal descriptivism.
To get a case where causal and deferential descriptivism really differ would require that 1) I intend to refer to the object which the person I heard the name from refered to, and 2) for some condition Phi, I regard my utterance of (*) as true iff the person who satisfies Phi discovered that dogs etc., and 3) Phi is not equivalent to the condition of being refered to by the person whom I heard the name "Oskar Minkowski" from.
It is hard to think of such a case. Whenever (2) and (3) obtain, a proponent of causal descriptivism could say that under these circumstances, I do not intend to refer to the object which the person I heard the name from refered to. Then causal descriptivism and deferential descriptivism coincide. But then "deferential descriptivism" seems to be the better name, as causal chains play no essential role in the account at all: all that matters is the conditions the person you've heard the name from had in mind.
Shifting the focus away from causal chains like this helps to handle the Madagaskar and the exchanged babies cases, where a name is introduced for one thing but then consistently used for something else. When Marco Polo used "Madagaskar", the condition Phi he associated with the name was arguably something like (the condition expressed by a suitable utterance of) "this island here which is called 'Madagaskar' by those natives": he probably wouldn't have judged all his writings about Madagaskar to be clearly false if the natives used "Madagaskar" for a part of the African mainland (as they did). So here we have a case where somebody doesn't clearly intend to co-refer with the people whom he acquired the name from. Cases like this will be very common if some mixed kind of descriptivism is right: if the conditions people associate with names usually contain some ordinary, qualitative conditions as well as deferential and meta-semantic conditions (and indexical conditions of personal acquaintance, in many cases).
On the other hand, I think deferential descriptivism (and with it causal descriptivism in its usual form) embeds an implausible kind of semantic voluntarism: that I can refer to whatever I want with my words.
The exam: In an exam on the history of physiology there's a question about the discovery of insulin. Thinking, confusedly, that my pet aardvark discovered the connection of insulin with diabetes in his experiments with dogs, I write (*). Then I continue, "Oskar Minkowski is still alive today, but he hasn't done any experiments in physiology recently".
It seems to me that the first sentence I wrote was clearly true and the second one false, no matter that I intended to refer to my pet aardvark. Again, what I meant by (*) was of course false, but it doesn't follow that by writing (*), I wrote something false. Rather, as I was mistaken about the conditions under which (*) is regarded as true, I wrote a true sentence wrongly believing that it expresses a belief which I wrongly believed to be true.
By contrast,
At the veterinarian: I bring my pet aardvark to the veterinarian in ordeer to get it vaccinated, introducing it as Oskar Minkowski. Asked how old the aardvark is, I say "Oskar Minkowski is about 6 years old now".
This time, what I said is true, and I was not mistaken about what the sentence means.
So the context of utterance plays an important role here that is left out in causal and deferential descriptivism.
Meta-semantic descriptivism is, I think, a step in the right direction. It only needs a little twist to build in the context-dependence: my utterance of (*) is true iff the contextually salient object called "Oskar Minkowski" discovered that dogs whose pancreas is removed develop the symptoms of diabetes. This addition is required anyway because most names are not unique.
By saying (*) out of the blue, accomodation ensures that the physiologist Oskar Minkowski is the most salient object called "Oskar Minkowski". -- Or maybe not quite accomodation as it is usually understood: what ensures that I refer to the physiologist is not that (*) is true of him, but that it is commonly believed to be true:
Rats: Oskar Minkowski, a German physiologist and brother of Hermann Minkowski, the inventor of the Minkowski space, discovered that the pancreas produces an 'anti-diabetic' substance later identified as insulin. But contrary to common opinion, he didn't make that discovery in experiments with dogs, but rather with rats. His publications say "dogs", but that's a printing error. Interestingly, more than 200 years earlier, a Russian monk also called "Oskar Minkowski" made the same discovery in experiments with dogs. But he didn't tell anyone about it and is since long forgotten. Much later, in 2006, I utter, out of the blue, (*).
In this case, I'd say that my utterance was false. So the utterance doesn't create a context in which it is true. But I think this is a common phenomenon about accomodation and nothing special about names.
What does it take to be called "Oskar Minkowski"? Not much. If I decide to give my pet aardvark that name, then it bears that name. Apart from being explicitly baptized "Oskar Minkowski", things can also get to have the name by satisfying conditions generally associated with it, as in the case of Madagaskar and the exchanged babies. And there might be further possibilities:
Swampman: By an incredible accident, several atoms in a swamp arrange themselves into an intrinsic duplicate of Donald Davidson as he was in 1976. The duplicate leaves the swamp and walks into the next city, where he introduces himself as Donald Davidson and, falsely believing that he has done enough philosophy in his life, becomes a carpenter.
The carpenter is clearly called "Donald Davidson", even when he first introduced himself with that name. But he was never baptized and he didn't at the time satisfy much of the conditions anyone associated with the name. (Or should we say that he baptized himself by introducing himself?)
I don't know what, in general, it takes to bear a certain name. But anyway, ignoring odd cases, we can say that according to meta-semantic descriptivism, you will regard my utterance of (*) as true iff some contextually salient object that was either baptized "Oskar Minkowski" or satisfies certain descriptions commonly associated with that name (in a contextually relevant community) discovered that dogs whose pancreas is removed develop the symptoms of diabetes.
Deference and referential intentions aren't explicitly mentioned here. That's how meta-semantic descriptivism can cope with The exam. In The pet aardvark, it could accept either judgement: maybe my strange beliefs suffice to make my pet aardvark the most salient Oskar Minkowski. But I think the more natural judgement from this point of view is that my utterance of (*) here is false. That's another advantage over deferential and causal descriptivism, as to me it is at least far from obvious that my utterance of (*) in The pet aardvark is true.
In a sense, meta-semantic descriptivism can accomodate both causal and deferential insights: if somebody was baptized "Oskar Minkowski" but no causal links lead from this event to the present use (like in Rats), then that somebody is often not a good candidate for being the contextually most salient bearer of the name.
So the meta-semantic account seems to be preferable to the deferential account, which seems to be preferable to the causal account. But as I said, I'd rather go for a cluster theory that adds in further elements. I should also, and rather urgently, say something about the status of the truth conditions I'm trying to spell out here: they don't appear to be conventional A-intensions. But fortunately, this entry is already far too long, so I'll just stop here.