Page 261 - Encyclopedia of Philosophy of Language

duced through the Manhattan Project. Having the same nature, then, is a matter of scientific discovery; internal structure is discovered only through empirical investigation. The link between referring and knowing is cut, for the introduction of the term does not depend on the term's coiner knowing the nature of this struc- ture, and hence the nature of the 'same kind as' relation. It follows that the users of a natural kind term need not, and often do not, know necessary and sufficient conditions for membership of the kind. Indeed they may not just be ignorant, they may be quite mistaken without prejudice to their capacity to use the name to refer to a kind. We have here not just an account of how kangaroos got their (English) name; it is an account of what reference is. On this view, the referential relation is just an appropri- ate type of causal chain between represented and representation.
4. CausalTheoriesandSemanticData
Causal theories of reference are partly motivated by the problems of semantic buck-passing. But they also help to meet problems for theoretical role views of reference. Reference often seems to be stable over major scientific change; Cuvier and Huxley both used 'species' to refer to species despite their work at opposite sides of a great scientific divide. However, because nothing fits the characterization that Cuvier gave of species—for example, he thought they were immutable—that poses a problem for theoretical role views and other related views of reference. We want to say that Cuvier was wrong in what he wrote about species, not that he wrote about, literally, nothing. Putnam's theory, in cutting the link between the capacity to refer and the capacity to characterize a kind correctly, explains how reference can be stable over theory change. Kripke took the rigidity of names, and the fact that our capacity to use a name is insen- sitive to our epistemic grip on its bearer, to count against description theories and in favor of causal ones. But Field and Devitt have emphasized the fact that causal theories might be ultimate theories of ref- erence. For causal relations between the speaking mind and bespoken world are physical facts.
4.1 Fregean Problemsand Description Theories
The ambitious program confronts two distinct classes of problem. It faces the naturalization problem; the fundamental semantic notion must be explicable non- semantically, otherwise there is an unexplained relation at the root of the theory (see Sect. 5). Further, though, and simultaneously, the theory must do jus- tice to the semantic phenomena. In the literature on names that has dominated discussions of reference, the semantic agenda has been developed around prob- lems associated with Frege. A theory of names, the idea has been, must do justice to the following facts:
(a) The sentence Mount Egmont = Mount Taranaki differs in some very important way from Mount Egmont = Mount Egmont even though Mount Egmont is Mount Taranaki.
(b) The sentence Alfred believes that Mount Egmont is the highest mountain in New Zea- land's north island differs in some important way from Alfred believes that Mount Taranaki is the highest mountain in New Zealand's north island, again, despite the identity.
(c) The sentence Vulcan is the closest planet to the sun is meaningful even though 'Vulcan' is an empty name.
Though they have received less attention, similar phenomena arise for general terms. Consider, for example, empty terms. Piltdown man was a fraud, hence the species Dawsoni never existed; still Eoan- thropus Dawsoni was the precursor to homo sapiens is meaningful, though false.
Description theories of names and general terms have many problems. It was on semantic grounds that Putnam rejected abbreviation accounts of general terms and Kripke descriptivist theories of names. But whatever their other troubles, description theories do
justice to these problems. If names are abbreviated descriptions, it is no surprise that distinct but coref- erential names are not semantically equivalent. If names are abbreviated descriptions, then a name is meaningful but empty if its associated description fails uniquely to designate some object. Russell's theory of descriptions thus solves the problem of the mean- ingfulness of empty names, granted an account of general terms.
The problem of belief contexts is tougher, but a common idea is that in such contexts it is not merely the reference of the name that is relevant to the truth- conditions of the belief sentences in which it appears; the mechanism of reference is also relevant. Frege christened mechanisms of reference 'sense'. Though 'Mount Egmont' and 'Mount Taranaki' refer to the same mountain, their mechanisms of reference, the descriptions associated with them, are not the same; hence the two belief sentences have distinct meanings.
Some at least of this machinery can be applied to general terms. These too have a mechanism of ref- erence of some kind. Though Weismann believed that the germ plasm was responsible for inherited simi- larities between organisms, he did not believe that DNA was responsible for those similarities, unaware as he was of the identity of germ plasm and DNA. Perhaps we can appeal to distinct mechanisms of ref- erence to explain the differences between these belief sentences, and the apparently related difference between Germ plasm is germ plasm and Germ plasm is DNA. Of course, within buck-passing semantics, this idea is of restricted utility; for it cannot be deployed for those general terms to whom the referential buck is ultimately passed. A logical empiricist might argue
Reference: Philosophical Issues
239