Putnam asks whether meanings are “in the head” and answers that they are not. More specifically, what he has to say about meaning and reference is that what we know of the meaning of a term will often be insufficient to determine what it refers to. Others have made similar points. In particular, Saul Kripke (who Putnam will refer to) has said something like this about proper names and suggested that natural-kind terms were similar. Putnam focuses on the natural-kind terms and considers a range of examples different in character from Kripke’s. I have outlined Kripke’s views below, and you should regard that account of them as part of your assignment; his views were well known and were being actively discussed at the time when Putnam published this paper.
As Putnam notes in his last paragraph, he sees two reasons why meanings aren’t in the head. After his introductory section (699-700), he considers a series of examples that illustrate one or both of these features of meaning (700-704). He then devotes a section each to discussing these two features, first what he calls “the division of linguistic labor” (704-706) and then what he calls “indexicality” (706-711). He gives more attention to the latter point; and it is what he says about this, along with the associated examples concerning “Twin Earth,” that has been most influential and that will probably occupy most of our discussion.
The paper is fairly short and Putnam writes clearly, but his examples and the points he makes about them are likely to take some time to digest, both in your reading and in our discussion. You should plan to have at least pp. 699-706—i.e., everything but the last section—read for Tuesday and the paper completed for Thursday.
Kripke’s own presentation of these ideas appears in an attack on Russell’s claim that proper names are really disguised definite descriptions (see, for example, The Problems of Philosophy, ch. 5, ¶¶17-24). Kripke’s attack was presented in a series of lectures in 1970 that are now published as Naming and Necessity (Harvard University Press, 1980). It has several components:
• Kripke says, first, that we often do not associate with proper names descriptions full enough to determine their reference (e.g., I might know that Millard Fillmore was an American president and be ready to use his name to make that claim without knowing even whether he preceded or followed Lincoln). More importantly, he holds that the descriptions we do associate with a name may not even fit the object named. If I associated with the name “Millard Fillmore” descriptions that happen to fit James Polk, I nonetheless have false beliefs about Fillmore rather than true beliefs about Polk. (I’m assuming here, not that I’ve mixed up the names but that my beliefs about Fillmore were the result of a number of independent mistakes and just happened to all apply to Polk.)
• He offers an alternative account of how proper names refer that has often been called a “causal theory.” A proper name begins to refer to a certain object when that object has this name attached to it—perhaps by pointing or perhaps by use of a description—and the name is then passed on to others who intend to use it with the same reference. Later users may never have been in a position to point to the thing named or have known the description used to “fix the reference” initially; and, over time as the name is passed on, false beliefs may come to be attached to it, and it could even be that few true beliefs remain. However, if there is an unbroken chain of people who each intend to refer to whatever was being referred to by whoever the name was learned from, then the reference is preserved in spite of the accretion of false belief.
• He holds that claims of identity between proper names are necessarily true if they are true at all. For, if N = M, then the names N and M refer to the same thing; and, to say that “N = M” might be false is to say that this thing might not have been itself. (If you hold that “a = a” can be true only if a exists, then what is necessary is the claim “If N, or M, exists, then N = M.”)
• Nonetheless, he says, such a claim might be known only a posteriori. For example, it is true that Hesperus = Phosphorus (these are names used for the evening and morning stars, respectively) since both name the planet Venus. But it must have required some degree of astronomical research to discover this truth (even though the discovery is probably too ancient to date). Analogously, Kripke says that “theoretical identities” between general terms in science (such as the identity of heat and molecular motion) are necessary truths known a posteriori.
• Kripke also holds that the are a priori truths that are not necessary, but only contingent. One example is “N is the so-and-so,” where “the so-and-so” stands for a contingent description used to fix the reference of the name N. For example, if the reference of a name N is fixed by the description “the object now in the center of this table,” the name remains attached to the object even when it is moved, so the description that is used to fix the reference initially is only a contingent property of the thing named. Still, at the time of naming, it is known a priori by the person naming the object that “N is the object now in the center of this table” because this holds, however briefly, simply because it was stipulated to be so. In short, Kripke holds that the two distinctions a priori vs. a posteriori and necessary vs. contingent are independent of one another, and he speaks of them as distinctions between “epistemic” necessity and contingency and between “metaphysical” necessity and contingency, respectively.