The Dispositional Convention Theory of Direct Reference for Names

Direct Name Reference Without Causal Links:

A Dispositional Convention Theory

John Dilworth, draft only, 11/04

A Kripke-style direct or causal theory of reference for proper names claims that three necessary conditions are involved: 1) A baptismal condition; 2) A linguistically normal causal chain condition; 3) An intention by succeeding speakers in the chain to use a name to refer to the same item as previous speakers in the chain. However, the first claim, as well as a translated form of the third, can consistently be held by descriptivists as well as direct reference theorists, and the second claim is false, as shown here with several counter-examples. I argue that direct reference theorists should instead endorse a replacement for thesis 2), namely: subsequent, non-baptismal uses of the name 'X' succeed in referring to X just in case the current use is a) a legitimate use of the name; b) one that uses the same name 'X' that was assigned to X in the baptismal ceremony.

A Kripke-style direct or causal theory of reference for proper names[1] may be approximately[2] specified in terms of the following three claims:

1) The name 'X' initially comes to refer to its bearer X in virtue of a baptismal ceremony, in which the name is assigned to X as part of the name-related linguistic practices of the relevant society.

2) Any subsequent, non-baptismal user Uj of the name 'X' must acquire the name in a normal linguistic way from some previous user Ui of the name, with this acquisition being one link in an historical chain of similar causal links, from user to user, which originates in such a baptism.

3) Subsequent to such an acquisition, user Uj succeeds in referring to object X by use of the name 'X' only if he intends to refer to the same object X as did user Ui.

Thus overall, Kripke requires as necessary conditions for a name 'X' to be successfully used by a speaker Uj to refer to object X a combination of baptism, normal linguistic causal links, and an appropriate referential intention on the part of the speaker Uj, who thus acquires the link from speaker Ui.

In most discussions of causal/direct reference versus descriptive theories of name reference, theses 1)--3) of such a Kripke-style direct view are discussed as if they are integral, or even inseparable, parts of a single direct reference thesis about name reference. However, the three theses not only can but must be separated, because thesis 2) concerning causal links can be shown to be false--as will be demonstrated here--while thesis 1) is, by itself, a generally acceptable thesis that descriptivists also could happily endorse.[3] In addition, a suitably translated form for thesis 3), purged of any mention of causally intermediary speakers--instead expressed in terms of a speaker intending to use 'X' to refer to the same object as did its baptismal originators--could also be accepted by descriptivists. Hence, in order to produce a philosophically interesting direct reference theory for names that is inconsistent with descriptivist views of non-baptismal singular reference, thesis 2) must be replaced by some other claim. One such claim may be introduced via the following discussion.

To begin, thesis 1) concerns the setting up of a social practice or convention, of referring to object X with the name 'X'. In the case of successful baptisms, issues specifically concerning singular reference to object X itself must have been internally resolved as part of the baptism, such as by the participants directly perceiving the object, or by using uniquely identifying descriptions in some such ways as are discussed by Kripke.[4]

Until those issues have been resolved, there is no object X to which a name 'X' has been attached. But once those issues have been resolved, there are no longer any active semantic, epistemic, metaphysical etc. issues as to whether, or how, name 'X' can be used by speakers to refer to X. The baptism has internally involved actual reference to X, and its output is a convention that the name 'X' may be, or is to be, used to refer to the same object X as was referred to by the baptismal participants. Thus what the baptism establishes is a practice or convention that name 'X' may be used to refer to object X in that society, by persons so intending to use it. Hence a more explicit analysis of Kripke's condition 1) will also absorb a translated form of the intentional condition 3), in that a proper or legitimate use of the name 'X' to refer to X will itself involve an intention by a speaker to refer, by means of use of the name 'X', to the same object X as was referred to by its baptizers.

Given these points, it would be a significant mistake to ask, considering subsequent, non-baptismal uses of the name 'X', under what conditions those uses of name 'X' really do refer to object X--rather than to some other object, or no object at all. Given the existence of the social practice or convention, any legitimate uses--conforming to the convention, and involving the relevant intention--of the relevant name 'X' by members of the society are guaranteed to refer to X. Hence it would also be a significant mistake to argue, as Kripke does, that we are able to succeed in referring to object X by use of name 'X' only because of a causal chain linking the current use to the baptismal use.[5] If the baptism genuinely did set up a convention that name 'X' refers to X, then no such causal chain could possibly be needed, since any legitimate, intention-involving and convention-conforming use of name 'X' will automatically refer to X, no matter how it came about that the user came to thus legitimately use the name 'X'. (Several alternative, non-causal-chain methods will be suggested later). Thus the current point counts both against Kripke's causal chain condition, and against any descriptivist substitute, neither of which are needed.

On the present account, there is only one other necessary condition--implicit in the above discussion--of successful reference to X in such cases, in addition to legitimacy of use of the name 'X'. The condition is that the relevant name must be distinct from other names that have the same linguistic expression in the relevant language. Thus for instance, if a user refers to a person George as 'George', that name must uniquely be a name of that particular person George, and not also of other persons whose names happen to be spelled in the same way.

Now it might be enquired whether Kripke's causal chain condition could be reinstated as one needed to guarantee sameness of a name 'X' between an original baptism of X and a later use of a name 'X'. However, the counter-examples to be given will equally count against that possibility. We can now introduce our thesis 2) which, as stated above, is a replacement for Kripke's causal chain thesis 2). For convenience it will be prefaced by the still-accepted thesis 1):

2) Subsequent, non-baptismal uses of the name 'X' succeed in referring to X just in case the current use is a) a legitimate use of the name involving an appropriate intention; b) one that uses the same name 'X' that was assigned to X in the baptismal ceremony.

The resulting pair of theses provides the basis for a convention theory of direct reference for names. It could be viewed as one logical extension of attempts, such as that of Kripke, to pare away all descriptivist elements whatsoever from cases of reference using names, since according to it successful referential uses of names need not even satisfy appropriate-causal-chain descriptive criteria.

Nevertheless, linguistic conventions cannot operate in a causal vacuum. It will be assumed that legitimate name uses at least have in common that they instantiate manifestations of appropriate, causally based linguistic dispositions. Thus the resulting direct reference theory could be called a dispositional convention (DC) theory of name reference. Also, such a theory can allow that the relevant dispositions are often produced by causal chains of a Kripkean kind, though they need not be, as the counter-examples to follow will show. Hence the DC theory involves at least dispositional causal elements, even though it is not strictly a causal theory of reference as usually understood, in that it rejects the causal-chain requirement of conventional causal theories.

1. Dispositional Reference for Names Via Duplication

The falsity of Kripke's causal chain thesis 2)--that subsequent, non-baptismal uses of the name 'X' succeed in referring to X only if the current use is a link in an historical chain of causal links originating in such a baptism--will now be shown with the help of a thought-experiment, which invokes the logical possibility of exact functional duplicates of persons, of a kind familiar from Davidson's 'swampman' example.[6]

But instead of just one person being duplicated, we may suppose that a whole society of people is so duplicated, so that questions of individual versus social reference remain undisturbed by the duplication. One possible scenario having that consequence is that, under threat of extinction within a few decades by a large asteroid approaching the earth, the current people all ship off to some more hospitable distant galaxy, leaving exact functional duplicates of themselves behind to populate the planet during its remaining years. The duplicate persons are such as might be produced by a 'star trek' transporter, that could make exact molecular copies of persons, so that issues of different kinds of causal instantiation of the same functions do not arise. Such duplicates have, of course, none of the causal history of the original people, and hence there are no relevant linguistic baptismal causal chains applying to them, that caused their current linguistic dispositions to be the same, or equivalent to, those of the original people. Specifically, such a duplicate never acquired any putative name in her vocabulary in a normal linguistic way from a previous speaker, and hence she also did not acquire an intention to to use the name in the same way as a previous speaker, hence violating both conditions 2) and 3) of Kripke's original set of conditions.

Nevertheless, my claim is that the duplicate people would have, not only exactly the same underlying causal dispositional powers as the original people,[7] but as a result of those dispositions also have, to a large extent, the same singular referential abilities as those of the departing people whom they duplicate. To be sure, the duplicates would have many false beliefs, especially concerning their relations to each other, unless the departing people had explained the situation. For example, if an original person Bill was married to Jane, their functional duplicates would falsely believe that they were married to each other. But the duplicates would have exactly the same linguistic skills and knowledge--including the same dispositions to use, for instance, the proper name 'the Parthenon' to refer to a unique building in Athens, Greece--as did the departing people.

Here again, the Parthenon-related beliefs of some duplicates may include some false beliefs, such as a belief by Bill's duplicate that he had seen the Parthenon previously, which corresponds via the functional duplication to a true belief of Bill concerning Bill's own actual prior visit. But as with causal theories of reference in general, the descriptive accuracy of the dispositional causal mechanisms that here provide sufficient conditions for reference by duplicates is not a relevant issue with respect to the referential status of their linguistic uses as such.

Now some points in defense of the above massive duplication scenario will be presented. To begin, to a large extent contentious issues about the cognitive contents of functional duplicates, as compared with those of the original persons, can be bypassed here.[8] If some wish to insist that, strictly speaking, the duplicates have no conscious beliefs, referential intentions and so on whatsoever, because of their lack of the right kinds of historical causation, then the point can be conceded for present purposes.[9] Arguably all that is needed for the relevant names to successfully refer is that the duplicates are indeed capable of carrying out substantially the same behavioral acts of singular linguistic reference, using those names, as could have been carried out by the original people, and doing so because of exactly the same internal causal dispositional powers possessed by those original people. Thus even if 'all is dark inside' the functional duplicates, still they could achieve reference for their name uses, learn it from others for reference to new particular objects, and so on, insofar as they contribute to a society whose overall linguistic sophistication and functionality is to a great extent continuous with that of the previous society of original people.

In support of this view, some cognitive concepts are relatively readily receptive or amenable to purely causal, non-experiential analyses. The suggestion being made here is that the most basic linguistic mechanisms of singular reference only require for their instantiation some relatively low level cognitive abilities, which could be possessed just as well by unconscious zombies--if such a concept is a consistent one--as by normal conscious people. All that is needed is some dispositional abilities to perceptually categorize situations as being of some particular type, plus some additional, socially supported dispositions to appropriate linguistic behavior depending on the results of such perceptions. Doubtless this is not sufficient for a full account of language understanding and use, but it surely suffices for legitimate competence in simple cases of singular referential linguistic use, which is all that is needed to explain referential uses of names.

In addition, another thought experiment will now be proposed, which suggests, against the above 'zombiehood of duplicates' assumption, that duplicate people would function exactly as we do, including having conscious experience. The thought experiment is a variant on the brain in a vat scenario, but with a crucial difference--namely, how do we know that we are not the functional duplicates of a previously departed group of people?

Such a scenario is crucially different from Putnam's original brain in a vat hypothesis, in which it is plausible to say that most of the beliefs of the envatted people are false.[10]

In the current duplication scenario, almost all of the duplicate people's beliefs--our beliefs--are true, except for a few isolated beliefs about our origins, which are matters of broadly religious or metaphysical controversy anyway. (This essay might indeed be, for all we know, the first to inadvertently reveal our actual origins as duplicates of departed people--unless reincarnation views count as preempting such a discovery.) Thus the issue is, how do we know that, at some arbitrary point in our history, all of the original people were not removed, and exact functional duplicates introduced in their place? Surely we don't know this not to be the case, even if it seems extremely unlikely. But equally surely, it should make no difference to our views about singular reference. Our continuing ability to make singular reference to Aristotle, for instance, cannot be a hostage to arbitrary historical episodes of people-duplication, whether or not such any such episodes have in fact occurred.

Or to put the matter in epistemic terms, we know that we are capable of singular reference to at least some historical figures, and certainly to many contemporary people and objects. But we know this independent of whether or not people-duplication episodes have occurred in the intervening years since the relevant items were baptized. Hence, since any such episode would destroy the postulated historical causal chains linking us to those historical figures or items, the existence of such complete, unbroken historical chains cannot be a necessary condition of our abilities to achieve singular reference to those actual items. Hence, in causal terms, only our current dispositions can secure such references for us--and the dispositional convention (DC) theory explains how that is possible.

2. Random Reference using Names

In this Section a much more pedestrian kind of counter-example to the Kripkean causal chain theory of names will be presented, involving no duplications of persons, or unusual counterfactual conditions of any kind. But arguably this very innocuousness makes such counter-examples much more difficult to dismiss than the previous more exotic cases.

To begin, recall that Kripke's theory claims that it is at least a necessary condition for a use of a name to refer to its bearer that its user should be normally connected to the relevant historical chain leading back to the initial baptism. However, consider the following kind of tactic, which doubtless has been used by investigators, scam artists and so on. The tactic is to associate a name with a person by some random technique, which likely often will fail, but which may also succeed occasionally. For example, if one systematically telephones every person in a large apartment block, and inquires of male respondents "Is this Paul?", it is likely that at least occasionally one would be successful, that is, one would have succeeded in referring to the person Paul, to whom one is talking, by use of his actual name 'Paul'. In such a case successful linguistic reference to a person X by use of his proper name 'X' is achieved without the random caller having had any contact whatsoever with causal chains originating from the baptism of that person Paul.

To be sure, the relevant investigator or scam artist in such a case knows nothing about the person to whom he is talking in such a case. But the case is a natural extension of cases such as that of Kripke's Feynman case, in which a person learns of the physicist Feynman through a legitimate historical chain, but who still may not know enough to identify him.[11] Thus, while Feynman was still alive such a person might have called every room in a conference that Feynman was attending, and asked of each male respondent "Is this Feynman?" Surely he would have succeeded in referring to Feynman by use of his name if in one such case his respondent had been Feynman himself.

Admittedly in such a case the enquirer did have a particular person in mind while calling, while the previous random caller did not--or at least, not while formulating the scheme. However, when e.g. the male in apartment 2a answers the phone, and the random caller enquires of him "Is this Paul?", the caller does have that particular person in mind as the person he is currently referring to as 'Paul'. But in any case, the issue being discussed is not primarily that of referential intentions of speakers--though see below for an invocation of such factors--but instead that of the referential success of their utterances. My main claim is only that such success can indeed be achieved by such random naming techniques, in spite of their extreme minimalism as compared to more central cases of referential linguistic competence by speakers using names.

Here is an explanation of what is occurring in such cases, in terms of thesis 2) of the current dispositional convention (DC) theory of name reference. There are two reasons as to why the random caller needs to actually contact persons, whether by phone, in person, and so on. The first reason is that our intuitive criteria for what counts as a legitimate use of a name includes a disposition to use it referringly, that is, such a disposition manifested with a particular person in mind. This the random caller achieves by getting some actual person on the line, so that it is him that the caller has in mind in exercising his disposition to call him 'Paul'.

And secondly, in purely dispositional or behavioral terms, the tokening of the word 'Paul' by the caller does not by itself qualify as a tokening of the of the particular actual name of some actual person, since there are many thousands of people called 'Paul'. Hence the context of use of the word 'Paul' has to be right for that use to count as a use of the name of a particular person. But any actual conversational interchange with a particular person, whom the caller refers to as 'Paul', would automatically count as a use of that particular person's name 'Paul', should the person actually be named 'Paul'. Hence both clauses of thesis 3, namely that the use is both a legitimate use of a name, and of the particular name of the relevant person, are satisfied by such a random naming episode.

The force of this kind of counter-example to Kripke's causal chain condition can be strengthened by noting that not only are some such chain-free examples possible, but that in principle the same random methods could be used to successfully refer to any object or person by use of its name. The possibility of such examples serves to emphasize what the current dispositional convention (DC) view of name reference insists on, namely that there cannot be anything more to successful non-baptismal reference to object X by use of its name 'X' than that the use is legitimate, and that the name 'X' is indeed the name of X. Hence the DC theory is indeed a satisfactory direct reference replacement for Kripke's descriptively hamstrung causal theory.

Notes