Abstract

Recently a fundamentally new picture of how names and natural kind terms refer has emerged. The "causal" theory of reference has illuminated the ways in which the connection between word and world depends not just upon the beliefs and representations of the speaker, but upon causal links between language users and their environment. This theory has drawn criticism on the grounds that it embraces an unacceptable metaphysics since it entails a form of essentialism. In my dissertation, I sort out different, logically independent components of the new theory of reference and contend that for natural kinds a metaphysically unobjectionable essentialism can be defended. According to this doctrine, the essences of paradigms of a kind determine the extension of a natural kind term, but the essential features are not fixed by a metaphysical distinction between necessary and accidental properties. ;This alternative theory of natural kinds is not original: Locke entertains it when he presents his theory of real essences, and Aristotle endorses it in his philosophy of science. Having articulated this alternative to the traditional account of natural kind terms, I consider which construal best accords with actual scientific practice. Elements of this benign essentialism appear in the writings of Galileo, Darwin, and Einstein. Further, the adoption of alternative semantic theories and alternative accounts of essences underlies both contemporary and historical debates in biological taxonomy and recent arguments concerning artificial intelligence. Finally, I examine how theories of reference naturally suggest methodologies of science, and explore the ramifications of this essentialism for accounts of scientific practice