Unlike my attack on Quine’s argument for natural kinds realism, my problems with Kripke’s natural kinds realism did make it into my dissertaion. My focus there wasn’t on natural kinds; it was, instead, on Kripke’s general theory of non-connotativity of certain terms (and natural kind terms are among them). But, at least for Kripke, non-connotativity of natural kind terms and natural kinds realism are closely tied, so the same argument will do for both. For those of you who aren’t terribly excited about the idea of looking up my dissertation, I’ll put a somewhat abridged version of my argument here.
Kripke’s Original Argument
Kripke’s argument for natural kinds realism is, essentially, this: We don’t specify, through anything like definition, the extensions of our natural kind terms. For example (his example), there’s no set of concepts associated with the term “tiger” that uniquely picks out all and only tigers. Indeed, tigers could turn out to have none of the attributes associated with our concepts of them. Although they, for example, appear to eat meat, be related to cats, and be striped, these could (though it’s exceptionally implausible–it’s not impossible, and that’s all he needs) turn out to have none of these attributes. The apparent relation to cats could be a result of our inadequate understanding of genetics, the stripedness could be an optical illusion, and the meat-eating could in fact be something very different (perhaps–remember, we’re allowing the implausible here–tigers’ claws convert the creatures they kill into 100% textured tofu, so they never consume an iota of meat).
(Kripke has another argument, about water, that goes in the other direction–that something can have all the properties we conceptually associate with a kind without being a member of the kind. I’ll cover that in a separate post.)
How do natural kind terms get their extensions? Well, Kripke wants to present, as opposed to a conceptual model, a causal model. The initial referent of “tiger” (or, more likely, some linguistic ancestor of “tiger”) was a tiger because of something that Kripke called a “baptismal event”–generally an ostention (“I shall call that thing [pointing] a tiger”) but possibly an *initial* description (“I shall call the creature that mauled this water buffalo a tiger”). At any rate, in all likielihood, this baptismal event picked out only a single object, or possibly a small number of objects. The relationship of that term to other tigers is that other tigers are part of the same natural kind. The extent of that natural kind is a scientific fact to be discovered, not a definitional one to be stipulated. And the relationship of our term “tiger” to tigers is that there’s the right sort of “causal chain” between the initial term’s usage and our current usage of “tiger.”
Now, Gareth Evans has a fairly famous argument against this, based not on a natural kind term, but a proper name, “Madagascar.” According to Evans (I haven’t checked his linguistic research), “Madagascar” was not, originally, the name of the island off of Africa’s east coast; rather, it was a name for a portion of the eastern African mainland. The use of the word to describe the island dates from European exploration of the region; Europeans misunderstood their African informants to be talking about the island when, in fact, they were not. Since the baptismal event of the term “Madagascar” (which well predated European exploration of the area) referred to the portion of the mainland, on a causal theory like Kripke’s, argued Evans, the referent of “Madagascar” should be a portion of mainland Africa. But, of course, it’s not–“Madagascar,” as it’s used today (even in Africa), clearly refers to the island.
Kripke’s response to this is that, while there’s a “causal chain” between the initial baptismal event and the term we use today, it’s not the right sort of causal chain. What’s the right sort of causal chain? Causal theorists of meaning are very vague on that point. Kripke, in particular, insists that what he presented is a “picture” of meaning, not a fully-fleshed out “theory,” and that as a “picture” it didn’t require a precise account of what sorts of causal chains are acceptable. More on that in a little bit–but what most causal theorists at least agree upon is that a miscommunication of the sort that happened with “Madagascar” is not something that could be a link in “the right sort” of chain, and this, I think, is fair enough.
However, I think it’s not too difficult to come up with examples of failures of the causal theory that do not involve bad links in this way. I’m not entirely sure that any of these examples are pieces of actual linguistic history–they are the sort of “common knowledge” stories that it’s extremely hard to find hard support for (I attempted to, when I was writing the dissertaion, to no real success), but I don’t think they need to be; it’s clear that they could be origin stories for natural kind terms, and that they wouldn’t (if true) have the effects the causal theory would predict.
One such example is a fairly common (though, again, rather hard to support) story about the origin of the term “unicorn.” According to some, at least, the term “unicorn” was involved in a baptysmal event with an oryx, a north African animal with two very even horns. The horns are so straight, and so even, that from the side (if one is of a credulous mindset), it’s not too difficult to believe that the creatures have just one horn in the middle of their heads. At any rate, the story goes, someone saw an oryx, falsely believed that it had only one horn, called it a “unicorn,” and went home to spread the tale. This story was passed down to us today, and is the source of our own “unicorn” term.
Now, if this causal chain is “the right sort” of causal chain, the causal theory should predict that the kind term “unicorn” refers to oryxes, or at least a particular type of oryx. Admittedly, oryxes don’t have many of the properties we associate with the term “unicorn,” but of course the whole point of Kripke’s tiger example is that that doesn’t matter–if the baptismal event is there, and the right sort of causal chain is there, the properties we associate with the term are entirely irrelevant. But, of course, “unicorn” doesn’t refer to oryxes. It’s an empty term, referring to nothing at all (or, if you prefer, a fictional term referring to fictional creatures). And this causal chain, unlike Evans’, doesn’t have any obvious candidates for “bad links.” There were no miscommunications, no obvious candidates for places where the meaning of the term might have changed. So this looks bad, or at least worse than Evans’ example, for the causal theory.
Is it doom for the causal theory? Well, not really. This might still, perhaps, be the “wrong sort” of causal chain–an advantage of a picture over a theory is that it always has an out with respect to any counterexample. But there are a couple of things worth noting:
- Whether or not this is a disproof of the causal theory, it’s pretty rough on, at least, Kripke’s “tiger” argument for it. (I’m going to address the previously mentioned “water” argument in a separate post, and look at Putnam’s independent arguments in yet another, but for now at least note this.)
- That “picture” stuff gets kind of old after a while. Not having an account of what precisely makes “the right sort” of causal chain was fine when Kripke published Naming and Necessity. But the causal account of reference has been dominant in the philosophy of language for 35 years now, and, to my knowledge, no real progress has been made towards specifying what “the right sort” of causal chain is. This ought to inspire some skepticism about the account’s outlook; if we can vitiate the arguments for the account, it ought to inspire a lot of skepticism.
To be continued…