Naming and Memory
Two theories of naming have dominated recent discussion: the description theory and the causal chain theory. But there could be others: descriptive fitting and causal connection are not the only conceivable relations that might underlie the naming relation. Earlier theorists might have suggested that a resemblance relation is the relation that underlies naming: the reference of a name on a given occasion of utterance is the object that resembles the mental image in the mind of the speaker. Such theories have not been popular since Wittgenstein’s criticism of image theories of meaning, and for good reason. But images do have referents, so they are the kind of thing that might logically qualify as determining the naming relation. Perhaps there are speakers elsewhere in the universe that invariably have detailed images corresponding to every name they use, and in fact these are the determinants of a name’s reference. This is the correct theory for them. We might want to modify it to accommodate the conceptual point that objects and images can’t really resemble each other, being entities of quite different types; but that is easily accomplished by saying that the image resembles the percept that speakers have in mind when they perceive the reference of the name. It appears that such a theory is pretty hopeless empirically for human name users, given their paucity of imagery, but in principle the theory could be correct for differently constituted beings. However, the theory suggests a wider range of options than is commonly recognized. Might there be a better theory that has some affinity with the image theory but avoids its pitfalls?
I want to suggest that memory provides such a theory. The basic idea is simple: the reference of a name is the individual the user of the name is remembering when he or she utters the name. More precisely, the user associates the name with a certain memory (possibly a memory image) of its referent. The name evokes a specific memory of particular person or thing, and it refers to the entity thus remembered. For example, I have memories of Saul Kripke derived from meeting him, and “Saul Kripke” refers for me to the person I am remembering when I use that name. So there are two relations that go into fixing the reference of a name: the memory relation itself (“m is a memory of object x”) and the association relation between a name and a particular memory (“m is associated with name n”). We need not go into what constitutes these two relations—it could be a causal relation and a relation of psychological association of ideas—what matters now is that the two together supply an alternative to the usual two theories of naming. In short: you are naming what you are remembering when you use the name. The remembering relation between memory image (trace, engram, etc.) and object is what underlies the naming relation. We are not limited to invoking descriptive relations and causal relations—that is, semantic fitting and social transmission. We need to consider the suggestion that the basic relation in naming is remembering x.
There is an immediate objection: what about naming things of which we have no memories? The objection must be conceded: we often refer to people and things of which we have no memory, that we have never seen, met, or experienced in any way. For example, I can refer to Plato (I just did) and yet I have no memories of Plato. But this is not a real problem for the theory, because we can simply take a leaf out of the chain-of-communication theory’s book: those with no memories of the bearer of the name refer to that individual by using the name with the intention of referring to the same individual as the speaker from whom they learned the name. So the theory is really a two-part theory: there are the in-the-know speakers with memories of the individual in question, and there are the speakers that are parasitic on these privileged speakers. This resembles the standard theory that combines an initial baptism with an historical chain of linked uses, but we substitute memory for baptism. Instead of saying that the reference of the name is fixed by a description or demonstrative in an initial baptismal act we say that speakers acquire memories of an individual and these memories fix the name’s reference.  Intuitively, you encounter someone, perceive that individual in some way, and form a memory of the individual in question; you then decide to call that individual by a certain name. To ascertain the reference of a speaker’s use of a name we need to know which individual is being remembered when the name is used. The mechanism of reference is the memory-name connection—who or what the accompanying memory image is a memory image of. The material of the memory trace can be of any type—sensory, linguistic, computational, analogue, digital, etc.—what matters is that it is a memory of a specific thing. Other speakers can then defer to these original speakers in their use of the name, relying upon their memory of the referent to gain referential traction. The original reference is fixed by something in the speaker’s mind but it isn’t a definite description or a conceptual content that uniquely individuates the reference; it is simply a memory of the object, whatever form that memory takes.
It might be countered that some names don’t rely on memory to achieve a referent: for example, we can just stipulate that the name “Albert” will refer to the first person born in the next century—and no one has a memory of such a future person. And does naming myself require that I have memories of myself? What about names of numbers? What about names of past objects that no one was around to remember? The answer to these natural questions is that no theory of naming should try to encompass every kind of name. There are different ways that names can hook up with objects: by descriptions, by demonstratives, by mental images, by memories, even by intellectual intuition. There is no such thing as the naming relation, if that means a single kind of grounding relation applicable in all cases. But the memory theory is a good empirical theory of most human names; it captures the most central cases, viz. our typical reference to people and places by means of ordinary proper names. The human institution of naming, as it now exists, is founded on human memory. If speakers were subject to widespread amnesia naming would not be possible in its current form: you have to remember the people you have met and you have to remember what names are associated with these remembered people. There doesn’t have to be a formal baptism for names to get introduced into the language, but there does have to be a general capacity to remember things. What all names do have in common is that they are a dependent mode of reference: they rely on other ways of singling objects out. This is not true for descriptions, demonstratives, images, perceptions, or memories; these don’t depend on some other type of reference to make them possible. But names have to piggyback on other referring devices, which can be of various kinds. We could justifiably speak of the “varieties of naming”. Still, for the vast majority of cases memory is central to our naming practices. We accordingly need to add the concept of memory to our account of names, as they mainly exist for us now, not just the concepts of description and referential link. The right final account probably includes all three elements suitably combined: memories, descriptive contents of memories, and interpersonal referential links.
The theory I have in mind combines features from both the classic description theory and the newfangled causal chain theory. Causality enters through memory itself as well as through the historical chain of uses; and the memory theory locates naming in a certain state of mind, viz. possession of a memory image. Memories are always partial and perspectival, like Fregean senses, so that aspect of the description theory is preserved—remembering Hesperus is not the same mental state as remembering Phosphorus. Memories don’t have reference by means of uniquely identifying descriptions, any more than perceptions do, but there is clearly a definite content embedded in a given memory. Hence the associated name can have sense as well as reference in virtue of these memories. Descriptions and demonstratives don’t invoke memory in this way: you don’t need to remember anything in order to employ these referential devices (except what their constituent words mean). But you can’t successfully use a name unless you either remember its bearer or are suitably connected to someone who does (for those names that actually do depend on memory). Naming is a bit like knowledge: knowledge too is either memory-based or testimony-based, with the latter radiating out from the former. Similarly, naming is either grounded in memory or it radiates out from that basis by mean of interpersonal linguistic links. We might call this “the extended memory theory if names” just to have a label. We often don’t remember people’s names, but we don’t typically forget name’s people: say the name and we reliably remember the person referred to. This is fortunate or else we would be unable to use names in the way we do. Naming and memory go hand in hand. 
 Note that babies are usually already named before a formal baptism is performed, so the baptism can’t be the mechanism whereby names are bestowed. The baptism is more a legal confirmation than an original source of naming.
 I haven’t discussed names for natural kinds as well as names for perceptible qualities like colors, but the same considerations apply mutatis mutandis to these. For example, the use of “red” to name the color red depends on our ability to remember what red is, as well as our ability to associate such memories with color words. In general we must not underestimate the role of memory in linguistic understanding and use.