A topic in the Open Knowledge Graph — a free, open map of 15,290 topics and the order to learn them in.
Terms like "water," "gold," and "tiger" rigidly refer to natural kinds defined by their microscopic essence rather than observable properties. Putnam's Twin Earth argument shows that the extension of natural-kind terms is not determined by psychological states alone—the actual nature of the world matters. This challenges internalist semantics and supports externalism about content.
Work through Twin Earth: on Earth, "water" refers to H2O; on Twin Earth, "water" refers to XYZ (chemically different but macroscopically identical). Despite identical psychology, the two speakers' terms have different extensions. Then generalize: for any natural-kind term, what determines its extension is the actual nature of the world, not internal cognitive states. Study how this applies to biological kinds and chemical substances.
From your study of Putnam's semantic externalism, you know the core thesis: meaning "ain't in the head." What a word refers to is not determined solely by the speaker's internal mental states — psychological duplicates can use terms with different extensions if their environments differ. Now let's examine *why* this is true for a specific class of terms: natural kind terms like "water," "gold," "tiger," and "elm." These terms share a distinctive semantic behavior that reveals how language latches onto the world at its joints.
When you use the word "water," you don't mean "the clear, drinkable, tasteless liquid" — or at least, that description doesn't *fix the extension* of the term. If it did, then "water is H2O" would be a merely definitional truth, and we couldn't discover the composition of water; we could only stipulate it. But we *did* discover that water is H2O — chemists found out something substantive about the world. Putnam's explanation: "water" rigidly refers to whatever natural kind has the same molecular structure as the paradigm samples we originally called "water." The microstructural essence (H2O) determines what counts as water in every possible world — even worlds where the macroscopic appearance differs.
This is the rigid reference feature of natural kind terms, parallel to Kripke's account of proper names. Just as "Aristotle" refers to the same person across all possible worlds (not whoever happens to satisfy the Aristotle-descriptions in that world), "water" refers to H2O across all possible worlds. The surface appearance — clear, drinkable, odorless — is how we *identify* water, but it is not what makes something water. Fool's gold (iron pyrite) looks like gold but isn't gold; this is possible precisely because "gold" refers to the underlying atomic structure (Au, atomic number 79), not the observable properties.
The philosophical implications cascade outward. First, a posteriori necessities: "Water is H2O" is necessarily true (water couldn't be anything other than H2O in any possible world) but is known empirically, not by conceptual analysis. This challenges the traditional equation of necessary truth with a priori knowability. Second, the division of linguistic labor: most English speakers don't know the atomic structure of gold, yet they use "gold" with the same extension as chemists do. The extension is fixed by experts and the causal-historical chain of use, not by each individual speaker's knowledge. You and a medieval peasant both say "gold" and refer to the same natural kind, even though the peasant has no concept of atomic number.
Natural kind terms contrast sharply with artifact terms (like "table" or "hammer") and functional terms. "Table" doesn't refer rigidly to some microstructural essence — tables are defined by their function and shape, not by what they're made of. An aluminum table and a wooden table are both tables; there's no "essence of tableness" at the microstructural level. This asymmetry between natural kinds and artifacts is itself philosophically significant: natural kinds are the ones where science can discover hidden essences, while artifact kinds are constituted by human purposes and practices.
Topics in reflective domains aren't scored by quiz answers. Read, reflect, and mark when you've thought it through.