The Causal Theory of Reference

 

 

Primary:               Putnam, H. (1975) 'The Meaning of 'Meaning''

                                Kripke S. (1972) ‘Naming and Necessity’.

 

Secondary:            Lycan, W. G. (2000) Philosophy of Language, London; Routledge.

                                Devitt, M. & K. Sterelny (1987) Language and Reality, Cambridge, MA; MIT Press

Evans, Gareth (1973) ‘The Causal Theory of Names’ in Schwatz, S. (ed.) (1977) Naming, Necessity, and Natural Kinds, Ithaca NY; Cornell UP, pp. 192-215.

 

Kripke's Theory

 

Recall that I said that there were two parts to Kripke’s theory of reference. The first part of this theory is a tale that he tells about what the semantic role of referring terms is supposed to be. What do they contribute to the meaning of the larger linguistic items in which they participate (principally sentences?) The second part, independent of the first, is a theory about just what it is that makes a word refer to the thing to which it refers. What is it – in the real world – that establishes a relationship between those two things; or, what is the nature of this relationship. I hope that you can see that the two types of question are really very different. If we’re just considering description theories, I grant that it’s easy to become a little bit confused about this distinction – as I think the early description theorists may have done – because (i) the role of the referring term in a sentence is just the role of the description, whereas (ii) the link between the name and the object said to be just that the object is the unique satisfier of that description; and so, the explanation is very similar in answer to both questions. We have to suppose that the answers would have to be very different in the case of a direct reference theory. In any case Kripke’s answer to this second question is fairly straightforward. Thus:

 

A name refers to something because there is a causal relationship between the use of the name and the thing to which it refers.

 

Suggestive Problems

 

Some such idea as this is suggested by two fundamental problems with the stories that we’ve been telling about reference so far.

 

a.                    Self-Containment of Descriptions

 

We’ve been looking in past lectures at various forms of theory of meaning for names, where the problem as we’ve conceived it is to understand what it is about names that regulates the way that they mean the things that they are supposed to mean. The description theory – in one way of understanding it anyway – claims that the meaning of a name is just the same as the meaning of some definite description, and the discussion has centred on the controversial consequence that any implication that can legitimately be drawn from sentences using names should be equally legitimate when some occurrences of those names are replaced by their corresponding definite descriptions. We think that that turns out not to be the case. But even if it did turn out to be the case, how far would that get us?

 

Consider what a description consists of. In the case of Aristotle, for example, the description would be ‘the teacher of Alexander’, plus a bunch of other stuff maybe. But this description appeals to terms like ‘teacher’ and ‘Alexander’ which are themselves in need of explanation. Perhaps ‘Alexander’can be equated to ‘the pupil of Aristotle’, but this would lead us to an obvious definitional loop – which is another problem you can chalk up to description theories it seems to me that there are likely to be lots of names that are only known with respect to these sorts of loops.

 

                Pegasus = winged horse that Bellerophon rode and lived on Mt Helicon

                Bellerophon = guy who rode Pegasus

                Mt Helicon = where Pegasus lived

 

Going back to the original problem, a bigger difficulty is trying to define ‘teacher’. Unless you’re going to define it as the thing which ‘Aristotle’ and ‘Stephen’ and … all other teachers have in common (also circular, this time in two different ways) you’re going to have to define it in other general terms, and you know – because your dictionary is of finite size, that somewhere this definition too is going to start getting circular. Somehow we have to break out of this circle, but the description theories give us no clue how to do this.

 

b.                    The Mystery of Senses

 

On the other hand, if the meaning of names is somehow to be explained in terms of sense, we have to have some idea of exactly what a sense is. Frege, as we saw, has a variety of different understandings of what sense is, but the principal concept is that it is a ‘mode of presentation’. OK. So what is meant by a ‘mode of presentation’?, and how is it that this adds something to the name that is distinct from its function of reference? We could suppose that the MOP is some type of abstraction, but this would make it still mysterious how the abstraction did its work.

 

What we really want, if we’re going to appeal to senses is some way of explaining them in terms of the other sorts of things that we accept exist in the world. We want, in short, to have a theory of meaning that appeals only to the sort of entities that exist in our scientific theories. Physical causes and histories are just the sort of thing that we’d be looking for, so a theory that claimed to be able to account for meaning in those terms is well worth looking at.

 

 

The Causal Theory of Reference

 

Any theory of reference of the kind that we’re now considering has to account for at least two things: in the first place, it has to account for the manner in which a name first becomes attached to an object; and, in the second place, it must be able to explain how it is that people not present at the naming and not acquainted with the object can use the name to refer to it. So let’s talk about that.

 

a.                    Baptism

 

The event which fixes the reference of a term – typically a new term – is called its baptism or a dubbing. The theory which tells us how the baptism/dubbing works is a theory of reference fixing. We can start by describing the standard form which such an event takes.

 

Let’s suppose that we want to name our dog ‘Fido’. In the presence of the object to be named, the person doing the naming points at the object and says, ‘this is called ‘Fido’’. We can see that in this case there is some sort of causal link between the object and the term. Just being present and observing the object involves a causal linkage between the object and the observer, and the observer seeing the object and pointing at it and saying ‘Fido’ is a chain of events distinguishable from other events in term of the causal links between the observer who makes this ostensive definition of the term and the object whose name is being defined, and the utterance of the name term itself. In this standard/normative case the introduction of the term is by ostension.

 

The baptism is not only effective for the person doing the actual pointing and naming. We have to imagine that anyone else who was also present at the dubbing event, and who has the required linguistic abilities, would also, in virtue of their observing that event, gain the ability to refer to the object using the name ‘Fido’. This too is to be understood in terms of the causal links that exist between the observer of that event and the participants in that event – she hears the word ‘Fido’, she sees the object Fido, she’s sees the pointing of the finger at Fido, and so on.

 

b.                    Reference Borrowing

 

But what about those people who did not observe the original event? Most of us here were not present when each of us were given our names (I mean Bob wasn’t there when Carol was named and Carol wasn’t there for Bob either) and yet we have no difficulty in using and understanding those names. What can account for that ability? Apparently, the story goes that those not at the original baptismal event obtain their ability to refer to the object using its name from other speakers who already possess this ability, and the ability of all speaker who use the name successfully can eventually be traced back to one of the participants in the dubbing. I can talk about Napoleon, because I have read about Napoleon in a book, and I’ve gotten my referring ability via that channel. The author of the book may have heard about Napoleon from a lecturer. The lecturer may have heard stories told by his relatives. The relatives will have heard from other folks. And some of those folks, somewhere along the line, wil have seen Napoleon being pointed to and addressed as Napoleon, and so on … all the way back to Mrs Buonaparte saying ‘What a lovely little boy. I’ll call him Napoleon.’

 

Each of those interactions is – we have to suppose – distinguishable as a causal relationship. In each case the causal linkage is such that the name use of the name is linked in the appropriate way to the object … and appropriate here just means that its linked in just the right way so that it does refer.

 

So here’s the story as Kripke describes it for a baby in N&N:

 

His parents call him by a certain name. They talk about him to their friends. Other people meet him. Through various sorts of talk the name is spread from link to link as if by a chain. A speaker who is on the far end of this chain, who has heard about, say Richard Feynman, in the market place or else where, may be referring to Richard Feynman even though he can’t remember from whom he first heard of Feynman or from whom he ever heard of Feynman. He knows that Feynman was a famous physicist. A certain passage of communication reaching ultimately to the man himself does reach the speaker. He then is referring to Feynman even though he can’t identify him uniquely.

 

You’ll notice here that there’s nothing required to be known about Feynman for the reference to succeed. The final speaker may have nothing but false beliefs about Feynman, but if the causal linkages that are in place are of the correct sort then he can successfully refer to Feynman. If a friend asks you ‘what is a manatee?’ and you answer by describing cattle in Bolivia, you not only give yourself much innocent pleasure but you also allow them to refer successfully to manatees in the future. Is that right, do you think? Or would a description such as that constitute a causally effective reference fixing process that would result in a redubbing of the manatee. We’ll get back to those sorts of questions a bit later.

 

Advantages of the Causal Theory

 

 

With any theory there are going to be pros and cons. We’ll start by having a look at the points in favour of this one.

 

a.                    Explains Sense

 

You’ll recall that we decided that Frege may well have been on the right track about how to solve the problems that arise for direct reference theories: that there is more to the meaning of a name- for example – than its referential function, there is also something like its sense. One of the major difficulties with the Fregean view of sense, however, was that we couldn’t think of any way of instantiating such a property in the real world. But now, if we take sense to be something like the ‘mode of presentation’ that Frege generally claimed it to be then we have a plausible candidate for that semantic object in the causal chain that is attached to each term and links it to the thing it references.

&nbs˙˙

i.                     You can see how this would work. Given two names that have the same reference like ‘Hesperus’ and ‘Phosphorus’ or ‘Tully’ and ‘Cicero’ we can distinguish the two by saying that the causal chain that links my use of the word ‘”Hesperus’ to Venus and my use of the word ‘Tully’ to the famous Roman orator are quite different from the causal chains that link my use of ‘Phosphorus’ to Venus and ‘Cicero’ to the orator. Thus, if the causal/historical links to the reference of a sentence (whatever that might be) is the compositional result of the combination of words in the sentence, then in sentences which differ only in that some of the component words have different causal/historical chains connecting them to their referents we can expect that there will be a corresponding difference.

 

ii.                    More than this, it looks as if the ‘mode of presentation’ is a pretty good description of what we actually think will turn out to play the essential role in these causal chains. It’s not likely that every causal linkage, however, indirect or contingent, is going to play some essential role in the causal/historical reference fixing procedure we’re talking about. It’s not likely that both myself and the dog being in sunlight, or my smelling the dog, or my hearing him bark, are the sorts of causal linkages that are going to be relevant to either dubbings or reference borrowings.

 

iii.                  We can also see that a causal/historical link of the right sort has other characteristics which were supposed to belong to the sense. For example, it was claimed that sense determines reference, and it is certainly the case that if you have a causal/historical chain that links the use of a word at one end of it, with an object at the other end, and this object is supposed to be the reference of the word, then there is a very definite sense in which the chain determines the referent of the word. It determines it just by having the word at one end and the object at the other.

 

iv.                  Finally, sense was supposed to be the essential thing for Frege in understanding the meaning of a word. It’s only if you have some special cognitive relation to this object ‘sense’ – which Frege called ‘grasping’ it – that you can be said to understand. Or rather, the very fact of your understanding is said to consist in your having this relationship with the sense. Well, as far as the causal/historical linkage goes, you can only be held to understand a word if your usage is properly placed in a causal chain linking word and object. It is fairly clear that some part of this chain is going to have to pass through the brain/mind of the language user. At various places in the explanation of how the causal/historical story of sense goes we’ve been mentioning as one of the qualifications the the user involved has to have a certain degree of linguistic sophistication. You can see why this is the case if you consider whether the dog that I baptize ‘Fido’ will perceive this as a baptism and will have the appropriate causal connection that will account for its use of the word ‘Fido’. I think not. Now, this ‘linguistic sophistication’ of which I speak,, which may also be thought of as language competence, is the sort of thing that results from causes in the brain. These causes are therefore essential to the story. The possession of particular causes of this sort by the word user can be thought of as constituting his ‘grasping’ of the sense. (It’s like he’s got hold of a chain somewhere in the middle links, in virtue of their passing through his head.)

 

b.                    Looks like language

 

Recall the things that we said were characteristics of language. This story about causal/historical chains allows us to account for a great many of them.

 

i.                     In the first place, we can see that the characteristic of stimulus independence is explicable, because whatever it is that leads to the emitting of a word like dog, it is not necessarily the presence of dogs or dog-like thing, or even reliable indicators of dog-presences. The causal chain  is sufficiently generally described that the very absence  of dogs could equally well be a stimulus for the use of the dog word (as in ‘where’s that darn dog when I need it.’)

 

ii.                    The characteristic of medium independence is explicable, because it really doesn’t matter in what ‘medium’ the word is transmitted in order for there to be an appropriate causal/historical chain linking the use of that word (in whatever medium) to the object that it’s supposed to refer to. I can write ‘come here, dog’ on a piece of paper and it will be ineffective as an actual command, but no competent language speaker will fail to grasp the meaning of it, nor will it fail to refer.

 

iii.                  The connection between word and object is arbitrary, just as we required, because the causal/historical connection does not depend upon the form of the utterance in any respect but only to the form of the dubbing ceremony or the reference borrowing processes, such that they provide the right causal/historical links in the chain. I can call all dogs ‘boggles’ for all the difference it’d make to the referring capacity of my language.

 

iv.                  The abstractness of names is explicable because the name does not depend upon any of the characteristics of the object in order to refer to it. Pictures, for example, don’t have this capacity because they rely upon a sufficient set of analogous structures in the representation to the object represented; and we might suspect that the description theory could also be seen as having difficulties in this respect. In the causal/historical story, there is no need to refer to properties of the object in order to construct an appropriate link between it and the word.

 

Recall that the other characteristics of language in general (as an abstract object) that we were interested in were power, productivity and systematicity. The causal/historical view doesn’t give us any real assistance here, but there don’t seem to be any reasons to think that the story is actually incompatible with languages possession of those characteristics. We may legitimately assume that they will be explained by some future refinement of the theory.

 

c.                    Solves the four puzzles of reference

 

We also need to be sure that the new theory is at least as capable as the Fregean theory or the Russellian of solving the sorts of problems that I have been taking as bellwethers for theories of reference or of language more generally.

 

i.                     We need to explain why it is that two identifying statements like

 

Venus is the evening star

Venus is Venus

 

can actually have different meanings or cognitive significances while the only difference between them is that different referring terms are used to refer to the same objects. You’ll remember that on the Millian View of direct reference – and possibly also on the Kripkean view – these got their meaning from the meaning of their parts, and the meaning of the parts ‘Venus’ and ‘evening star’ was no more than their reference, and their reference is identical, and so the two sentences should mean the same.

 

The causal/historical story also takes reference to be essential to the meaning of referential terms, but the fact of these terms referring is explained in terms of their linkage to the objects by some chain of causes. This chain can be used to distinguish two different terms that refer to the same object, just as sense was supposed to do for Frege. The link from ‘Venus’ to Venus is different from the link from ‘the evening star’ to Venus, therefore the person who understands the first sentence has to be able to cope with two different causal chains connected to Venus, whereas to understand the second sentence he only needs to be able to cope with one. (Note that I’m not claiming that he has to be able to cope with the causal chain connected to Venus in order to determine whether the second sentence is true or not. That doesn’t require any understanding of the referring term at all.)

 

ii.                    We have to be able to explain how it is that we can make meaningful statements that appear to refer to objects that don’t exist. How is it that we can say something like

 

Pegasus had a glossy white coat

 

And be in no danger of saying something meaningless? If the meaning of a sentence is derived from the meaning of its part, and one of those parts has no meaning because it has no reference, then the sentence should be meaningless.

 

In this sort of case the causal/historical theory claims that although there is no actual object to which a causal chain from ‘Pegasus’ can be attached (in the way that causal chains need to be attached in order for them to refer) this does not mean that there is no a highly analogous causal chain attached that can play much of the role of sense-giver that the designational causal chains play for actually referring terms. Consider how a typical such causal chain arises: someone dubs a thing by a name and then by the use of that name in conversation and writing and so on a lot of causal links arise that allow others to become part of the causal chain that eventually can be traced back to the object named. In the case of these fictional names the name is introduced to conversation by some fabulist and lacks an object which is to be the subject of a dubbing, but the rest of the causal connections deriving from the conversations and writings that use that name are exactly the same as in the standard case – and therefore is claimed to play the role of sense equally effectively.

 

You may wonder whether this should be taken to suggest the idea of a chain fixed at one end only (by the term) and with the other end thus free to flail about doing all sorts of referential damage. This is not necessarily the case. Consider how the name ‘Peasus’ is introduced: someone says ‘There is a wonderful creature that is just like a horse but it has wings and can fly.’ In such an introduction the storyteller is giving a description which has a meaning and a sense because the referring terms in it, ie. ‘horse’, ‘wings’, do refer – though they refer in this case to abstractions, which is another story. In short, it’s because of this grounding of the introductory description that ‘Pegasus’ can be said to be referentially fixed.

 

Just note also that this story gives a good reason for distinguishing between names that are fictionally based like ‘Pegasus’ and names that are just nonsense syllables like the ‘Jabberwock’ or ‘borogoves’ in the Lewis Carroll poem. Names like those don’t seem to be introduced by descriptions and so they don’t have any grounding at all and so you’d probably claim that they didn’t have meanings/senses. Does this mean that sentences that use them are meaningless? Do you think that the lines

 

        Beware the Jubjub bird, and shun

        The frumious Bandersnatch

 

Are meaningless?

 

iii.                  We have to be able to explain how it is that we can make meaningful statements that appear to deny that something exists. You’ll recall the problem there for direct reference theories: if your statement denying that Pegasus exists is true then the statement refers to a non-existent thing so there is a failure of reference so there is a failure of meaning so the statement is meaningless. It can only be meaningful if it is false. Inconceivable!

 

But the same story that explains why reference to nonexistent fictional creatures is meaningful will suffice to explain why existence statements that refer to them will be meaningful. The nonexistent thing being referred to is not at the root of the causal chain attached to the term ‘Pegasus’, but there is a chain and it is grounded and so there is a sense-analogue for the term and so it can be said to be meaningful and so the sentences truly denying its existence can be meaningful. They are meaningful in just the same way that sentences falsely denying the existence of me are meaningful.

 

iv.                  The final problem that we need to be able to deal with is the problem of reference in opaque contexts. It’s unfortunate that this turns out to be a fairly tricky problem for causal/historical theories. Those who are interested in seeing how this turns out could have a look at Michael Devitt’s (1981) book Designation where he has a go at working out what’s going on.

 

 

Problems for the Causal Theory

 

 

I’ve done enough advertising for this theory now. It’s time to have a look at some of the problems that people have thought that such theories face. A lot of these problems were brought forward by Gareth Evans – not the former Australian foreign minister and present head of the International Crisis Group, but a quite different Australian. (Same name, different referent. How does the causal/historical theory handle something like that?)

 

a.                    Change of reference

 

The first problem to look at is the problem of change of reference. We are all aware of the fact that words can begin by referring to one thing and finish up referring to some other thing. The example that Evans gives is a report that

 

“In the case of ‘Madagascar’ a hearsay report of Malay or Arab sailors misunderstood by Marco Polo … has had the effect of transferring a corrupt form of the name of a portion of the African mainland to the great African island.”

 

He also has a simpler example that we can use:

 

Two babies are born, and their mothers bestow names upon them. A nurse inadvertently switches them and the error is never discovered. It will henceforth undeniably be the case that the man universally known as ‘Jack’ is so called because a woman dubbed some other baby with the name.

 

The problem is that the dubbing is pointing at one object originally and yet through various historical events it seems that the reference has switched to some other object. How can this be the case if the reference is determined by the original dubbing event and this original dubbing event is not changed – in the way that all past events can’t be changed.

 

The standard response to a difficulty such as this is to suppose that the causal/historical chain that connects a name to its bearer is actually a whole lot of chains: the claim would then be that there are typically several chains that are connected to the object that reach to the term that names it. You can see why that would be a plausible claim to make because it’s not usually the case that an object is pointed at and named on just one occasion and all other references are purely derivative of that event. Consider this, for example: Aurelia, who’s one of the lecturers here, has just had a baby; she brought the kid in to show us how well she’d done, and while she was here she was asked what it’s name was several times, and each time she answered ‘He’s called Guy.’ It seems to me that each of those episodes is going to be at least as referentially significant as the original dubbing event – certainly they’re going to be more referentially significant to anyone who was not present at the original event. They play, for me anyway, the role of dubbing event. All my subsequent referrings to this baby – though there will not be many – will take their referential power from the causal chain that is grounded with Aurelia’s statement in the presence of myself plus the baby.

 

This idea of multiple grounding of reference in the object then allows an explanation to be given of reference change such as we had in those two examples. How so? Because it allows for there to be groundings that are inconsistent with an original dubbing. The baby Jack in Evans’s example is confused with the baby Jim. The original referential role of the dubbing event upon the other baby – let’s call it Jim – is transmitted through just one causal chain, but the large number of subsequent grounding of the name Jack in that baby are able to overwhelm the referential significance of that original dubbing is that the baby can really be said to have been named Jack.

 

Something similar happens with the Madagascar example. In that case however you may find that there are two communities of speakers who are using the words with different causal chains attached to them. One bunch, the original users of the term are using it to refer to part of the African mainland, while another bunch, who got their use through reading Marco Polo, are referring to the island. Who is right in this case? Couldn’t they both be right? There may be a period of confusion if the two groups ever had occasion to talk to each other about Madagascar, but this would be exactly what we should expect. We really do have experiences of such confusions, so it’s a point in favour of the causal/historical view that it tells us how such things are possible and how they come to be resolved..

 

b.                    The quâ-problem

 

Another difficulty, which is not so easily gotten around, goes by the name of the quâ-problem. Consider this: I point at an object, for example, a dog, and I say ‘Lo! This is called Fido’. I then take it that I have named this object, this dog, Fido. But what is it that distinguishes this naming event from the naming event where I point at a dog and say ‘Lo! This is a called a dog.’ How does it turn out in the first case that the causal anchor of the referential chain is laid in the particular dog I called Fido, whereas in the second case the causally relevant division of the world is all things that are dogs? How, indeed, did it turn out that in the first instance I was taken to be pointing at the whole object – the dog qua dog – and not at the dog’s nose and had thus named the dog’s nose ‘Fido’, or perhaps the wetness of the left nostril, or perhaps a time slice of the dog’s passage through the space-time continuum, or etc.?

 

Now, it is a fact that names don’t typically point at any of those other things, but it is another fact that they could. We could very easily come up with names for such things and ground them in what would apparently be the very same ‘object’. This sort of thing strongly suggests that there is something about the grounder himself that is relevant to what is really the object that is being baptized. The obvious thing that could be different in the person doing the grounding and responsible for the type of grounding that occurs, is the thing that we are sure is different in the grounder; that is, the intention of the person doing the grounding, by which I mean their mental state of intending to refer to the dog  rather than to time slices of the critter.

 

What this means is that there has to be something like a description involved in the baptismal event. A description is required to make the claim that one has an intention to point at an object of a particular kind rather than of some other kind. It is not necessarily a uniquely identifying description of the object in the sense that the descriptions proposed by Russell and Searle might have been, but it does mean that we can’t really think of the causal/historical theory as completely unrelated to the description theories that we are intending to replace. There is some degree of descriptivism required to make it work.

 

Natural Kind Terms

 

 

There are plenty of other possible problems that can be found with the causal/historical view of reference, but those two will do for the time being. Let’s take a look now at yet another extension of the basic causal theory.

 

We’ve now seen a few cases in which the causal/historical story relies upon an ability to name more general classes of objects. When we were looking at the talk about fictional objects like Pegasus we supposed that the original grounding involved descriptions of wings and horses and not ostensions of any particular wing or horse. Similarly, we have just seen that the solution to the qua-problem for the theory involves some sort of descriptive element in which the kinds of things being referred to are specified. The natural next step for the causal/historical theory would therefore seem to be to see if it can be applied to the naming of natural kinds. The sorts of natural kinds that are used in this discussion are those that were used by Kripke and Putnam (in your readings); thus we’ll be talking about gold, water, and tigers.

 

a.                    Problems for Description Theories of Natural Kind Terms

 

Let’s remind ourselves that the theory that Kripke and Putnam were trying to replace was basically the descriptivist theory, and the story that that theory told about the meaning of natural kind terms was that it was given by a description or a description-cluster of the properties of the natural kind. Thus gold, say, was to be described as a a heavy, yellow, noble metal; and water is a colourless, odourless, potable liquid; and a tiger is a wild, striped, feline. Kripke and Putnam point out that there are very cogent criticisms against this view of the meaning of natural kind terms – in fact they’re pretty much the same type of criticism as they made against the descriptivist theory of names. In particular they make modal arguments against all such descriptions: for any such description we can conceive the situation that (ie. there is a possible world in which) the description fails for members of that natural kind. For example, gold could have not been yellow, water could have not been odorless, and tigers could have been only striped because the freemasons had painted them that way for their own nefarious purposes. There are also problems with unexpectedly trivial statements like ‘water is potable’, which turns out to be tautologous in this case.

 

These are serious difficulties indeed, but the most serious difficulties – possibly fatal difficulties – arise when we consider that the descriptivist theory of natural kind terms, like that for names, requires a good deal of knowledge on the part of the user for their intention to refer to a natural kind to be achieved by their using the name of that kind. Putnam makes the point that he cannot reliably distinguish an elm tree from a beech tree because he does not have, either explicitly or implicitly, the knowledge in the form of a description that would suffice to identify either natural kind. Yet when he says that elms are trees or when he hears others talking about elms in autumn leaf he is surely referring to elm trees and he is surely being told about elm trees. Thus identifying knowledge is not necessary for successful reference and to grasp the meaning of a term.

 

On the other hand, identifying knowledge is not sufficient either. Devitt and Sterelny give the example of the discovery of a creature that satisfies all the identifying descriptions of Pegasus. It’s a real flying horse. Perhaps it was bred by the Freemasons in a secret gene splicing factory in Tibet., If so, it is not the Pegasus to which we were referring. (And this is not necessarily because our description includes the word ‘mythological’, other non-mythological examples could be given. The point is just that no reasonable description is going to complete enough to specify only the thing to which we wish to refer and no other possible object. Or kind)

 

This is an appropriate point at which to introduce you to the famous TwinEarth argument by Putnam. It may be taken as an argument against the possible adequacy of descriptive theories in the original Russellian sense. Imagine that there is a world far far away from Earth, but a world in the same universe as Earth – not a merely ‘possible’ world then – and this world is a near duplicate of our own familiar planet. A casual observer would see the same plants and animals and seas and continents, and even the same people speaking the same sentences in the same language. In fact, however, TwinEarth differs from Earth in that whereas water on Earth is made of H2O, what the inhabitants of TwinEarth call ‘water’ is made of XYZ. Now, XYZ is a colourless, odourless, potable liquid and has all the other properties that a seventeenth century Englishman would have given to the thing he called ‘water’. But ‘water’ despite the identity of the descriptions does not mean the same thing on TwinEarth as it does on Earth, because it points to XYZ rather than H2O. This indicates the failure of the description theory to determine reference properly. It also indicates, that the descriptions as instantiated in someone’s head, their thoughts or intentions, can’t be identically referring either; and therefore, because the heads of the speakers are otherwise identical, it appears that meaning cannot be defined entirely in terms that do not reach beyond the mind of the speaker. As Putnam puts it, meanings just ‘ain’t in the head’.

 

b.                    The Extended Causal Theory

 

So what is it that Kripke and Putnam say makes water water, gold gold, and tigers tigers? Well, they claim, it is just the scientific natures of those things. In the case of water, it is the fact that it is composed of H2O; in the case of gold it is the fact of its being 79Au; in the case of tigers, the fact of their having a particular genetic coding. But this, you may object, hardly seems likely; after all, couldn’t it be the case that in some possible world water was other than H2O, gold other that 79Au, tigers other than the very genetically definite creatures that they are? Kripke and Putnam say no. There is no possibility of water being other than it is; there is only the possibility of one’s being ignorant of what it is.

 

The reference fixing part of the extended causal theory appeals to this claim of direct reference by Kripke and Putnam in order to establish that there is a real sense in which there is an unambiguous natural kind at the base of the chain. The chain for ‘water’ is established by pointing at several instances of water. It then turns out that the causal chain is grounded in (refers to) the natural kind formed by the scientific description of the attribute that defines the similarity of all those examples that were pointed at. (Note that the science that lies behind the naming of natural kinds is just the same science that uses those natural kinds to establish its laws, so the process is not so unlikely as it might at first appear.)

 

The rest of the causal/historical story is much the same as for naming particulars, so we don’t need to repeat it all here. You may also wonder whether the problems that were suggested to exist for the singular version of the causal/historical theory could not be repeated for this more extended version. I’ll leave that for your own consideration.