Name-bearing, reference, and circularity

Proponents of the predicate view of names explain the reference of an occurrence of a name N by invoking the property of bearing N. They avoid the charge that this view involves a vicious circularity by claiming that bearing N is not itself to be understood in terms of the reference of actual or possible occurrences of N. I argue that this approach is fundamentally mistaken. The phenomenon of ‘reference transfer’ shows that an individual can come to bear a name in virtue of the referential practices of a group of speakers. I develop a picture of name-bearing which captures this fact by treating the extension of name as a function of the way that extension is represented in the presuppositions of groups of speakers. I show that though there is a form of circularity inherent in this approach, it is not vicious circularity.

Here, the expression Alfred 1 expresses a metalinguistic property-something like bearing the name 'Alfred'. Properties like this are not, in general, uniquely instantiated. All Alfreds are in the extension of Alfred, all Helens are in the extension of Helen, etc.
The central feature of PV is the claim that this predicative metalinguistic meaning is the meaning of a proper name; that is, the same expression that occurs in constructions where a proper name has the overt syntactic place of a predicate, as in (1), also occurs where names appear to be complete noun phrases-as in (2) 2 (2) Alfred is in the kitchen baking pies.
In (2), Alfred appears to function as a semantically simple referring expression (call this a bare occurrence of a proper name). PV holds that it is a metalinguistic predicate; how could that be?
Here the morphosyntactic component of the view comes in. The account holds that what appear to be simple referential occurrences of proper names hide semantic complexity. According to the simplest version of the view, bare occurrences of proper names involve an unpronounced definite determiner. The idea is that the semantic interpretation of (2) involves a structure like: (3) the null Alfred is in the kitchen baking pies Here the null Alfred is a definite description like any other-except that the definite article is not pronounced (i.e. null). So PV treats the semantic contribution of a bare occurrence of a proper name as semantically complex-as involving the metalinguistic predicate and some term-forming operator. 3 the null Alfred in (2) refers to the salient individual who bears the name 'Alfred' in the same way that the dean in (4) The dean is in the kitchen baking pies. refers to the salient dean. Support for PV is thought to come from a range of different sources. In addition to the fact that proper names can occur in the overt syntactic place of a predicateas in (1), above-there is a range of relevant cross-linguistic evidence. There are languages in which proper names do not have bare occurrences; that is, there are languages where names always occur as a constituent of a complex noun phrase. In some such languages, a typical use of a proper names in reference involves a definite determiner-for example, in some dialects of German you don't see Hans but der Hans. 4 The thought is that the simplest way to deal with this range of cases is to treat names as predicates and hold that languages differ with respect to the availability of the null form of the definite article in constructions involving names.
Proponents of PV also claim that there are interpretations of bare occurrences of names which the orthodox approach-which treats bare names as individual constants-cannot accommodate. A central case here are examples of covarying interpretations of proper names, as in (5). (5) If a child is christened 'Bambi', then Disney will sue Bambi's parents. (Geurts 1997, p. 321) 5 Here the bare occurrence of Bambi does not refer to some particular Bambi. The sentence is naturally interpreted as making a claim about different Bambis relative to different hypothetical christenings. PV is well-suited to explaining the availability of this sort of reading given its treatment of bare occurrences of names as definite descriptions. 6 Finally, PV holds out hope for addressing some of the traditional philosophical puzzles surrounding the orthodox view of names. To take one example, the orthodox approach does not sit comfortably with the meaningfulness of vacuous names. But if proper names express properties, and bare occurrences of names are interpreted as definite descriptions, it is no more mysterious that there are meaningful occurrences of, say, Pegasus, than it is that there are meaningful occurrences of the winged horse.
PV has been defended in a number of works-see Bach (1981Bach ( , 2002, Geurts (1997Geurts ( , 1999 and Elbourne (2005, Chap. 6). 7 My goal here is not to advocate for the 4 For the claim about German dialects see Elbourne (2005, p. 173). For other discussions of the relevance of cross-linguistic evidence to PV see Larson and Segal (1995, p. 355), Matushansky (2006Matushansky ( , 2008 and Ghomeshi and Massam (2009). 5 For more putative examples see Elbourne (2005, Chap. 6). 6 For what it's worth, it's not clear that examples like this should be taken at face value. For an extended discussion of this issue, see Gray (2012, Chap. 2) and Hawthorne and Manley (2012, Chap. 6). 7 For other discussions of PV, see Loar (1976Loar ( , 1980, Katz (1977Katz ( , 1990Katz ( , 2001, Devitt (1980), Abbott (2002), Elugardo (2002), Rothschild (2007), and Maier (2009). I ignore, for the sake of simplicity, a class of related views. Francois Recanati (1997, Chaps. 8, 9) defends indexicalism about names. According to that view, proper names are indexicals. The character of a proper name N is a function from a context, to the contextually salient individual who bears the name N (this view is elaborated and defended in Pelczar and Rainsbury (1998) and Pelczar (2001)). I would reference,and circularity 209 view, but rather to show how it ought to be elaborated. Much of the focus of these works has been a defense of the idea that bare occurrences of proper names are (at the level of semantic interpretation) definite descriptions. The central question is whether PV can do a better job than the orthodox account of names of predicting and explaining the range of potential interpretations of bare occurrences of proper names. The subject of this paper, in contrast, is a relatively neglected aspect of PV. The literature to date has tended to avoid serious discussion of the nature of the properties expressed by nominal predicates. In this paper, I will argue that the little that proponents of PV have said about name-bearing properties is misguided. I hope to show that PV is committed to a kind of circularity in its picture of the meaning of nominal predicates, but also that this circularity is benign. The structure of the paper is as follows. In Sect. 1, I introduce the main theoretical challenge for an account of name-bearing in the context of PV: the circularity challenge. In Sect. 2, I outline and criticize the standard response to the circularity challenge, noting that it fails to account for one of the two central ways that an individual can come to bear a name. In Sect. 3, I introduce an approach which is in a position to show what those two ways-baptism and reference transfer-have in common. In Sect. 4 I elaborate the approach by showing that while it is not viciously circular, it does commit PV to a kind of virtuous practical circularity in its conception of name-using practices. In Sect. 5, I defend the analysis by elaborating the conception of a name-using practice implicit in it.

Circularity
Anyone who has recently advocated for PV has felt the need to respond to an influential argument in Kripke (1980). Kripke proposes the following condition on theorizing about names For any successful theory, the account must not be circular. The properties which are used in [the determination of what object a name refers to] must not themselves involve the notion of reference in a way that is ultimately impossible to eliminate (1980, p. 68).
As an example of a theory of names which violates this principle, Kripke cites a theory proposed by William Kneale according to which '' 'Socrates' just means 'the man called 'Socrates' '' (Kripke 1980, p. 68 quoting Kneale 1960.

Kripke continues
As a theory of the reference of the name 'Socrates' it will lead immediately to a vicious circle. If one was determining the referent of a name like 'Glunk' to Footnote 7 continued class these views, along side the predicate view, as broadly metalinguistic views of names. What is characteristic of metalinguistic views is the idea that the property of bearing N is semantically involved in bare occurrences of N. Different accounts of this semantic involvement amount to different kinds of metalinguistic view. Although I won't argue for it here, my hunch is that the dialectic I develop in this paper with respect to PV, suitably modified, would apply equally to any metalinguistic view. himself and made the following decision, ''I shall use the term 'Glunk' to refer to the man that I call 'Glunk' '', this would get one nowhere. One had better have some independent determination of the referent of 'Glunk'. This is a good example of a blatantly circular determination.
Kripke's opponent distinguishes two (putatively) distinct kinds of facts: namebearing facts and name-reference facts. The proposal is that name-reference facts are grounded in name-bearing facts. The fact that Glunk refers to a is explained by the fact that a is called 'Glunk'. Kripke's complaint seems to be that name-bearing facts themselves are grounded in name-reference facts. What could it be for a to be called 'Glunk' if not for Glunk to refer to a, or for speakers to use Glunk to refer to a, or for speakers to take it to be proper to use Glunk to refer to a, etc.? What else might being called 'Glunk' consist in?
There are two aspects to the response from proponents of PV. Note that the proposal Kripke is attacking here is not precisely PV. He seems to have in mind a view which treats names as simple referential terms whose reference is fixed by a uniquely identifying descriptive condition (''the man that I call 'Glunk' ''). But PV holds that names are predicates, not singular terms, and they are typically true of many different individuals. PV holds that bare occurrences of names (the sort of occurrence that Kripke had in mind) are definite descriptions. Given that many different individuals typically satisfy a name, bare uses of names will typically be incomplete definite descriptions. This means that the name does not carry the full weight of achieving reference 8 -as Burge says, bare occurrences ''rel[y] on extrasentential action or context to pick out a particular' ' (1973, p. 432). 9 So the first part of the response is to point out that PV does not claim to be a full account of how the reference of a bare occurrence of a name is achieved in context. And typically proponents of PV are happy to accept a broadly Kripkean, i.e. causal-historical, account of reference-determination. That a given bare use of Aristotle refers to the philosopher and not to the shipping magnate is determined by causal historical relations holding between that use and the philosopher. PV is an account of the meaning of names, not an account of reference. 10 But the circularity challenge can also be applied directly to PV's account of the meaning of names. It might be that PV's appeal to name-bearing properties, as the meaning of nominal predicates, is somehow internally inconsistent. One might hold 8 ''One should also note, of course, that in the current view names contain a free variable which can be assigned a referent directly. So the descriptive content does not in any case bear the whole burden of getting reference off the ground.'' (Elbourne 2005, p. 177). 9 Burge holds that bare occurrences are semantically equivalent to demonstratives. Contemporary proponents of PV hold that they are incomplete definite descriptions. Depending on their view of definite descriptions, contemporary proponents differ with respect how contextual factors have their effect. Bach (2002) treats definite descriptions as Russellian and so treats the reference of a incomplete definite description as a matter for pragmatics. Elbourne (2005) treats definite description as complex individualdenoting expressions which contain variables. In the referential use of an incomplete definite description, the variable is free and assigned an individual by context (in this sense, his picture is broadly analogous to Burge's). Geurts (1997) employs an account of definite descriptions in a dynamic-semantics framework. 10 Geurts says that Kripke's complaint ''presupposes that any semantical theory of names should be a theory of reference'' (Geurts 1997, p. 325). Name-bearing, reference, and circularity 211 that those properties could not have the structure that PV claims them to have. Consider the relationship between the predicate Alfred and a bare occurrence of it (which, recall, is treated as a definite description of the form the null Alfred). PV concedes that part of what explains the reference of a given bare occurrence of Alfred is the extension of that predicate (just, for example, as part of what explains the reference of an occurrence of the dean is the extension of the predicate dean).
And PV holds that the extension of Alfred is determined by distribution of the property of being called 'Alfred'. But if we hold to Kripke's thought that namebearing facts are determined by name-reference facts we are faced with circularity. The reference of a bare occurrence of Alfred is partially explained in terms of the extension of Alfred which in turn is determined by facts about the actual or possible reference of bare occurrences of Alfred. At this stage the worry is still of the hand-waving variety. I will attempt to make it more precise shortly, but proponents of PV typically don't bother. They reject it out of hand. The second part of the typical response to the circularity objection is to categorically deny any substantive connection between reference and namebearing. 11 The response to Kripke on the part of PV has been to deny that one needs to invoke any notion of reference to make sense of name-bearing. Geurts and Bach, respectively, write 12 : Bearing a name is like wearing a tie. Like ties, names are seldom unique, but circumstances permitting they may be used for referential purposes. More accurately, just as you can employ the attribute of wearing a tie to identify to your audience the person you have in mind (John, as the case may be), you can use the attribute of being named 'John' for the same purpose. Taken on its own, however, a name doesn't refer any more than a tie does (Geurts 1997, p. 326).
It is no more essential to the property of bearing a certain name that one be referred to by that name than it is essential to the property of having a certain social security number that one be referred to by that number (Bach 2002, p. 83).
Here both writers are explicitly responding to the circularity worry. Their response is to claim that there is no worrying connection between bearing N and being referred to with N. Of course one can use the predicate Alfred to refer to someone who bears that name-for example, by employing a definite description of the form the null Alfred. But one can just as well use the predicate mayor to refer to someone who is a mayor-by employing a definite description like the mayor. It is no more essential to being an Alfred that one is referred to with the null Alfred than it is to being a mayor that one is referred to with the mayor. Or so the thought goes.
I am going to argue that the though letter of Geurts' and Bach's claim is correct, the spirit is decidedly on the wrong track. In order to side-step Kripke's circularity worry, they focus on cases in which an individual comes to bear a name in virtue of some baptism or stipulation. While such cases clearly exist-and are probably the most common-they are not the only way an individual can come to bear a name. There are also cases in which an individual comes to bear a name in virtue of the referential practices of a group of speakers-so called 'reference transfer'. 13 The standard picture of name-bearing associated with PV cannot capture these cases precisely because they reveal a dimension in which facts about name-bearing depend on facts about reference.

Name-bearing according to the predicate view
In what sense is Geurts' and Bach's view correct? Bach claims that ''[i]t is no more essential to the property of bearing a certain name that one be referred to by that name than it is essential to the property of having a certain social security number that one be referred to by that number.'' I take it he has in mind something like the following: an individual can bear a name N without ever having been referred to with N.
And surely one can bear a name without ever being referred to with it. An individual can some to bear a name in virtue of a properly situated stipulative act. A child is born, and baptised as 'Alfred'. Shortly after that, the world comes to an end. The child bore the name 'Alfred' yet was never referred to with Alfred. 14 So let's take it as a condition of adequacy on our account of name-bearing that it can allow for, and hopefully illuminate, the fact that an individual can come to bear a name without ever having been referred to by that name.
The letter of Bach's and Geurts' claims is correct. To see what is wrong with the spirit, we can look at what, beyond the denial that name-bearing stands in any constitutive relation to reference, is involved in their picture of name-bearing. In discussing circularity, Geurts writes: In my native country, last names are assigned according to strict regulations, but first names are afflicted by whim. Other cultures make use of patronymics, and in still others parents are named after a child. And so on. What all these naming practices have in common is just that some association is established between a name and its bearers, but how this association is initiated and sustained is different from case to case. The name-bearing relation between 'Lolita' and the famous novel was initiated by the author and is sustained, inter alia, by printing the name on the front cover of every copy. The association between my last name and myself is sustained, inter alia, by records at a register office, but this does not apply for my first name. If I 13 I say 'so-called' because if PV is correct, these cases do not involve reference transfer but rather involve a change in the extension of a nominal predicate which is not precipitated by baptism or stipulation. 14 Don't be tempted to say: wasn't the person in question referred to with the name in being baptised? Without going into the semantics of nominative constructions-although this, in itself, an interesting question (see Matushansky 2008)-we can note that one can be baptised by describing the new name: ''I give him his father's name'' or ''I give him the name which is spelled 'A' _ 'l' _ 'f' _ 'r' _ 'e' _ 'd'' ',reference,and circularity 213 decided that I wanted to be called 'Rudolf' instead of 'Bart', it would just be a nuisance to the people in my social sphere, but I don't think it would be humanly possible to change my last name into 'Carnap' (not in my country anyway). (Geurts 1997, p. 327) Here Geurts expresses the dominant picture associated with PV. The guiding thought-motivated by an attempt to avoid Kripke's challenge-is that the practice of naming has no constitutive connection with reference. Proponents of PV invoke broadly social and psychological practices associated with names: dubbings, birthcertificates, religious ceremonies, formal introductions, patronymics, nick-names, diminutives, etc. The picture seems to be that name-bearing comprises a motley of different social/psychological relations in which one might stand to a name. One might bear a name in virtue of it having been written on an official piece of paper on the occasion of your birth, or in virtue of it having been uttered by someone who was attending to you under certain canonical conditions, or in virtue of your friends calling you to mind when they hear the name spoken, etc. This picture is fundamentally misguided. I'll start by noting an odd feature: there doesn't appear to by any unifying principle which explains what baptism, birthcertificates, nick-names, etc. have in common. Geurts claims that ''[t]he expression 'bearing a name' covers as many relations as there are naming practices'' (1997, p. 328). This would be helpful if we had some sense of what sort of practice counted as a naming practice. With regard to that question he writes ''[w]hat all these naming practices have in common is just that some association is established between a name and its bearers, but how this association is initiated and sustained is different from case to case'' (1997, p. 327). This is clearly not going to help. There are any number of associations that might be established between myself and a name; most of them do not result in my coming to bear that name. To take an example: 'Alvin' is my favourite name, and all of my friends know that. Whenever they hear the name 'Alvin' they call me to mind. This is an association which has been established between me and 'Alvin'. It has a range of social and psychological features. But this doesn't mean that I bear the name 'Alvin'. Not just any association that is established between a name and an individual is a name-instituting association.
Perhaps this is not in itself objectionable. Perhaps there is no reductive analysis of name-bearing in terms of naming practices. The appeal to naming-practices is just supposed to show how we can get some grip on name-bearing without invoking reference. But even this seems implausible. What is a baptism? Clearly there are various cultural dimensions to baptism, but it is hard to avoid the thought that a baptism canonizes a way of referring to the baptised individual. Baptising an individual with the name N, I take it, is instituting a practice of using N to refer to that individual. Exactly what form that practice takes might be shaped by cultural idiosyncrasies, but reference is at the core. Imagine that we came across an undiscovered community, and some practice of theirs which involved uttering a name N over a newborn infant under particular conditions. What would convince us that this was a practice of baptising the child with N? I take it that if, going forward, N was used to refer to that child-even if, say, only during religious ceremonies or until a certain age-that would be a good sign that the child had been baptised. If, going forward, the father of the child was referred to with N (when he had not been before) we might think that this community had a practice of baptising fathers after the birth of a child. If the ceremony has no appreciable effect on the practice of using N to refer to anyone, we would at least be inclined to think that what we witnessed was not a baptism at all. Similar remarks apply to other naming practices. Birth certificates, for example, are taken to be part of a naming practice because they establish rules regarding how individuals must be referred to in official documents, etc.
Geurts tries to avoid the thought that name-bearing has any essential connection to reference by appealing to practices like baptism, birth certificates, etc. It is not clear, however, that we will be able to make sense of those practices without appealing to a notion of reference. So it's not clear that in appealing to namingpractices Geurts has ultimately avoided connecting name-bearing and reference.
I will offer a final, more substantive, consideration against Bach's and Geurts' approach-the moral of which will lead us toward a positive account of namebearing. The island of Madagascar apparently acquired its name as the result of a confusion with a certain part of the African mainland which bore that name (or, rather, a similar one) (Evans 1985, p. 11). Given that it was a confusion, the island's coming to bear that name involved no stipulation, agreement, or baptism. Some initial confusion-or even some group of independent confusions-brought about a stable practice on the part of some speakers to use Madagascar when they intend to refer to Madagascar and to take their cohorts' use of Madagascar to indicate an intention to speak about Madagascar. For some initial segment of the practice of using Madagascar to refer to the island this practice was based on an error about the relation between 'Madagascar' and Madagascar-speakers falsely believed that Madagascar bore the name 'Madagascar'. At some later stage, and without imagining any baptism or any other canonization, the stability of the practice of using Madagascar to refer to Madagascar becomes inconsistent with the idea of an error on the part of the speakers. That practice of reference itself results in the island coming to bear the name 'Madagascar'.
What such cases show is that there is a special form of dependence of namebearing facts on facts about reference. An individual can come to bear a name in virtue of the referential practices of a group of speakers. Proponents of PV claim that there is no more essential connection between name-bearing and reference than there is between, say, tie-wearing and reference. Their idea was that though it is certainly true that speakers can make reference to someone by using a nominal predicate they satisfy, they can also make reference to someone by using a description of the tie they are wearing. The claim was that Alfred doesn't stand in some constitutive relation to reference that, say, wearing a red striped tie does not. But clearly this cannot be right. Consider an analogous scenario to the case of Madagascar, but involving ties rather than names: we can imagine some general shared hallucination to the effect that a particular man, a, always wears a red-striped tie; and further, that a stable practice of using the man wearing the red striped tie to refer to a develops (that is speakers use the man wearing the red striped tie when they intend to refer to a and take their cohorts' use of the man wearing the red striped tie to indicate an intention to speak about a). I take it that there is absolutely no pressure to conclude that by virtue of such a practice a comes to satisfy wearing a red striped tie. Facts about tie-wearing do not depend on the referential practices of a community; facts about name-bearing do.
So we have a second condition on the adequacy of our account: our account of name-bearing should allow for, and hopefully illuminate, the fact that an individual can come to bear a name in the absence of any baptism or stipulation, in virtue of the referential practices of a group of speakers.

Name-bearing and presupposition
Our picture of name-bearing should illuminate two simple facts: (1) an individual can come to bear a name without having been referred to it with it (2) an individual can come to bear a name in virtue of the referential practices of a group of speakers. Worries about circularity have led proponents of PV to develop theories which only have a place for (1). But it is relatively straightforward to show how a theory can illuminate both facts, and to show that it is not thereby viciously circular.
What we need is a picture which unifies cases of baptism and reference transfer; we need a picture of name-bearing which reveals why these are two different ways of coming to have the same property and how each achieves that effect. The basic idea will be this: in both cases, what is created is a certain stable set of dispositions in a group of speakers. And it is these dispositions which constitute name-bearing. The challenge is to characterize what sort of dispositions are at issue, and how they are created by baptism and reference transfer.
To find the right set of dispositions, we should look more closely at the way that predicates get involved in reference-our task is to bring name-bearing and reference closer together while still allowing for the possibility that an individual can bear a name without ever having been referred to with that name. The characteristic way in which predicates become involved in reference is by occurring in a definite noun phrase-for example, a definite description (the F), complex demonstrative (that F), or possessive (John's F). A shared characteristic of definite noun phrases is that in uttering them, speakers typically presuppose that some satisfier of the predicate is uniquely identifiable by the participants in the discourse. This can take two forms. Either it can be presupposed that there is a unique satisfier of the predicate relative to the contextual domain. Or it can be presupposed of some individual x that x satisfies the predicate and is uniquely salient in doing so. In this second sort of case, the predicate is suitable, in the context, for reference to the individual. 15 The speaker's use of the predicate as part of a definite noun phrase of the right form will successfully indicate his intention to refer to the individual in question.
The presuppositions in a context are a function of the propositional attitudes of the participants. 16 The presuppositions in a context are the information against the background of which participants speak and interpret the speech of others. Roughly, a proposition P is presupposed in a context iff speakers are willing to behave for the sake of communication as if it is common knowledge that P. It is important that we look at propositions which speakers are willing to behave as if they believe, rather than propositions they actually believe. A speaker need not believe that O is drinking a martini to use the man drinking the martini to refer to O, she must be willing to act as if she does (perhaps because she takes it that her audience does in fact believe it). This is to say that there is an essentially practical element to presuppositions. Presuppositions do not directly reflect what participants believe, but how they are disposed to act in a communicative exchange.
My suggestion is that we should understand name-bearing in terms what is typically presupposed about the extensions of nominal predicates. This will connect name-bearing and reference (because which predicates are suitable for reference to which individuals in a context is a function of the presuppositions in that context) without entailing that to bear a name an individual must have been referred to by it (because presuppositions are dispositions to behave in certain ways, not actual speech behaviour).
What presuppositions are relevant? We have at least two options here. Consider a small village with two barrel-makers. Everyone knows each is a cooper, everyone knows everyone else knows this, and so on. So in typical contexts involving members of the town, it will be presupposed that each satisfies cooper. But in many contexts both will be equally salient, so it will not be presupposed that either is the uniquely identifiable satisfier of cooper. And so cooper is not suitable, on its own, for reference to either in such contexts. Contrast this with the village blacksmith. There is only one blacksmith in the village. In typical contexts involving members of the village it is presupposed that he satisfies smith and is unique in doing so among salient referential targets. So speakers can use smith to refer to him.
There are two levels of referential potential here. At one level, we have typical presupposition of satisfaction of a predicate; at the other, we have typical presupposition of unique satisfaction relative to contextually salient referential targets. Which sort of potential is relevant to name-bearing? It is difficult to give a principled reason here before looking more closely at the idea of a group of speakers operating in our characterization of name-bearing. But I can offer a preliminary reason to prefer the weaker sort of potential. To bear a name, it is not required that the name typically be suitable, all by itself, for reference to an individual (even amongst a group of speakers who know its name). It seems plain that we can imagine a community of speakers in which two different people, who tended to be referentially salient in most of the same contexts, bore the same name, say 'John'. In typical contexts, then, speakers could not use John, by itself, to make reference to either. But this does not, I take it, put any pressure on the idea that each is called 'John'. 17 Adopting the weaker kind of potential for our account, we propose that the following clause gives the meaning of the nominal predicate Alfred: (6) x satisfies Alfred iff there is group of speakers such that in typical contexts involving members of that group it is presupposed that x satisfies Alfred.
This approach has the virtue of being able to accommodate the fact that an individual can bear a name without ever having been referred to with it-because what is at issue is not how speakers have used the name but how they are willing to use it (though exactly how baptism tends to bring about the right set of attitudes remains to be shown). It also respects the moral of the story of Madagascar; the example of reference transfer showed that what is distinctive about the properties expressed by nominal predicates was that objects came to bear them in virtue of the way that speakers used those predicates in reference. Rather than looking at the use itself, we are now looking at one aspect of the source of that use (the stable presuppositions of speakers).

Circularity redux
I'm going to argue for the general approach developed in last section by showing how it can unify and illuminate baptism and reference transfer. I will work my way towards doing that by revisiting the question of circularity.
Recall that PV's response to the circularity objection has two parts. The first part is to note that PV treats bare occurrences of names as incomplete definite descriptions. So the extension of Aristotle does not bear the full weight of explaining how a particular bare occurrence of Aristotle might refer to the philosopher rather than the shipping magnate. Here PV is happy to take on board the causal theory of reference and so hold that a bare occurrence of Aristotle might refer to the philosopher in virtue of a chain of communicative acts which has its source in that individual. We should pause to assure ourselves that this is consistent with the account developed above.
Kripke held that the links in the causal chain which connect a particular utterance of a name to its referent are constituted by speakers' intentions. When a speaker, A, first encounters the use of the name Aristotle to refer to the ancient philosopher, she may form an intention to use the name to refer to the same individual (1980, p. 96). If a later use of Aristotle, by A, stands in the right relation to that intention, then that 17 An issue that would have to be addressed in a fuller account is the interpretation of compound names, part of the point of which, presumably, is to achieve reference in cases of this sort. The predicate view is better positioned to understand the (quasi) compositional interpretation of compound names than is the orthodox approach. But the issue is not straightforward. In particular, it's not obvious that we can understand the mode of combination of a compound name like Paul Ryan on the model of predicate composition. Doing so would seem to predict that Paul Ryan was synonymous with Ryan Paul-which doesn't seem right. use refers to the ancient philosopher. 18 Is there anything about this story that PV cannot simply co-opt? It appears not. The difference is simply that the relevant acts of reference involve complex referential terms containing predicates, rather than semantically-simple referential terms. When first encountering the use of Aristotle to refer to the philosopher, a speaker may intend to use that predicate to refer to the same individual. If later deployments of that predicate stand in the right relation to that original intention, that use of Aristotle will likewise refer to the philosopher.
It would be possible to hold that the causal-historical picture of referencedetermination essentially depends on the links in the chain each involving the deployment of a semantically-simple referential term in a public language. This would make PV incompatible with the Kripkean picture of reference. I'm not aware of anyone who takes this position, nor of any good reason to hold it. More than that, it seems unpromising on its face. The causal picture is also supposed to capture how the causal-informational processes in perception fix the reference of perceptual states (and the thought and talk based on them). 19 But those causal links do not involve the deployment of a semantically-simple referential term in a public language, so clearly that is not required in all reference-fixing causal chains. 20 So there is no challenge, as far as I can see, in simply adapting this aspect of the orthodoxy. In this I am simply following previous proponents of PV: Bach writes, ''The historical chains of reference that Kripke speaks of are perfectly real, but what these do is connect acts of reference made by speakers back to the individuals named'' (2002, p. 98 note 26).
One important note: (6) appeals to the de re presuppositions of groups of speakers (i.e. presuppositions of the form x satisfies Alfred rather than of the form The F satisfies Alfred). Thus the account assumes that an individual can come to bear a name only in virtue of the communicative practice of speakers who adopt such presuppositions about it (this assumption could be dropped, see note 22). 21 An issue in the background here is exactly what is required for a thinker to be able to frame de re presuppositions about an individual (or more generally, frame de re thoughts about an individual). On traditional accounts, a causal-informational connection to the object is required to frame such thoughts. And like the account of reference-fixing above, I see no reason that PV cannot simply adopt this aspect of 18 I make no attempt, and neither did Kripke, to specify what the 'right' relation is. It is surely not necessary that a speaker is able to recall her first encounter with the name and intend to co-refer with that initial encounter. More likely, her current intentions and beliefs with respect to the name must stand in the proper causal connection to the original intention, and she must intend to use the name in the way that her peers do. 19 See, e.g., Grice and White (1961), Evans (1982, Chap. 6), Récanati (2012, p. 56), Hawthorne and Manley (2012, Chap. 1). 20 For a negative assessment of the idea that semantically simple referential terms play an essential role in reference-fixing causal relations, see Goodman (ms). 21 There would be a form of vicious circularity present if one held that an individual might come to bear a name in virtue of presuppositions of the form 'The unique Alfred satisfies Alfred'. This presupposition could only concern a particular individual if that individual were already an Alfred. So the existence of such a presupposition could not explain that individual's coming to be an Alfred. But the above account is not committed to coherence of this possibility, so no trouble arises. Thanks to an anonymous reviewer for raising this issue. Name-bearing, reference, and circularity 219 the orthodoxy if that is desired. The natural thing to say is that speakers inherit the de re presuppositions of those from whom they acquire the use of the name. 22 The more serious circularity challenge concerned PV's account of the connection between the meaning of nominal predicates and the reference of bare occurrences containing them. Geurts and Bach claim that name-bearing has no essential connection to reference. They do this in order to avoid the charge that by explaining the reference of some particular occurrence of a name N by appealing to the property of bearing N, they are involved in vicious circularity. They are strictly speaking correct: name-bearing is not essentially connected to reference. But it is essentially connected to a particular range of speakers' attitudes with respect to names-namely presuppositions-and these in turn partly determine referential behaviour. So it's not clear that we really have avoided vicious circularity. To see this, note that (6) represents the extension of Alfred as a function of speakers' typical attitudes about the extension of Alfred. This feature of the approach captures the response-dependent character of name-bearing properties. But it has the look of circularity, so we should assure ourselves that nothing has gone wrong.
We should try to be more explicit about what would make some characterization of the meaning of a predicate viciously circular. 23 The most obvious form of vicious 22 I'm being a little cagey about this simply because I'm not confident that the traditional picture of de re thought is correct. It is possible to hold apart the question of linguistic reference-fixation from questions about the transmission of de re thought. It would be possible to hold that the conditions under which a speaker can achieve linguistic reference to an individual by deploying a name are more permissive than the conditions under which a speaker can frame de re thoughts about an individual (this is essentially Evans' position in (1982, Chap. 11) as well as the position advocated in Goodman (ms)). It also possible to be more liberal about the conditions for de rethought. One might hold that a causal-informational connection to an object is not necessary, and that is possible to frame a de re thought by, say, applying a dthat operator in thought to a definite description (Jeshion 2002) (see also Hawthorne and Manley (2012, Chaps. 1-3) for a good overview of the state of the art). Though I take no stand on the issue, it does affect the position developed here in subtle ways. In particular, if we want to hold on to the possibility of (socalled) descriptive names (Let's call the shortest spy 'Boris'), and we want to hold on to the idea that only de re presuppositions are relevant to name-bearing, then we have to allow that speakers can form de re presuppositions without a causal-informational connection to an object. Otherwise we would have to lift the requirement that only de re presuppositions count towards establishing the extension of a nominal predicate. This would require reworking (6).
I will also mention that if one is tempted to understand de re thoughts in terms of 'mental names' (i.e. simple referential terms) in a language of thought, PV is no obstacle to doing that. PV makes no claims about thought, its only target is natural-language syntax and semantics. 23 Though there is not space to fully address it there, there is another sort of circularity worry one might have about PV. The discussion in the body of the paper focuses on whether our attempt to characterize the extension of Alfred somehow made illicit reference to that extension. But one might think that there is an even earlier worry: (6) makes no reference to the extension of Alfred, but it makes reference to Alfred itself. Isn't this illegitimate? Can an expression be referred to in characterizing its own meaning?
There is much to say here, and much depends on one's views about the metaphysics of expressions. I'll simply note that this structure is not unheard of. Say one thought of the meaning of a variable as a function from an assignment to an individual in the domain. One might also think of an assignment as a set of pairs of variables and objects from the domain. In that case, one would be characterizing the meaning of an expression in terms of (an object which constitutively involves) itself. The account of nominal predicates developed here is similar.
Of course one need not think of an assignment in that way-one might think of an assignment simply as a sequence of objects from the domain. To do that, though, one would need to posit more structure in circularity occurs when an expression is used in its own definition. 24 (6) does not have this form; in it the defined expression is mentioned on both sides of the definition. But this, by itself, does not mean that the definition is not viciously circular. The definiens is meant to characterize a function which determines the extension of the definiendum. If the putative function explicitly or implicitly makes reference to the extension of the definiendum, the definition will be viciously circular-presupposing what it tries to establish. The most obvious case of this sort would be a characterization as in (7) x satisfies Alfred iff x satisfies Alfred in which the extension of Alfred is represented as a function of itself. But we could also imagine a hidden form of this circularity in which, perhaps through a series of definitions leading back to it, the extension of a predicate is represented as a function of itself.
(6) is not guilty of this; it does not represent the extension of Alfred as a function of the extension of Alfred; rather it represents the extension of Alfred as a function of the way that Alfred is represented in the presuppositions of contexts of utterance. And how Alfred is represented in the presuppositions of a context is not a function of the extension of Alfred. In general, x's being in the extension of a predicate P is neither necessary nor sufficient for it being part of the presupposition of some context (or some range of contexts) that x is in the extension of P (so there is no hidden chain of functional dependence).
This sort of functional circularity is only one sign of vicious circularity. Kripke's gloss-in relation to the example of Glunk-was epistemological. Another way of thinking about what is vicious in a viciously circular definition is that it initiates an epistemic regress when employed in an attempt to establish the applicability of the definiendum to an object. Burgess (2008, p. 220), following Humberstone (1997), gives a characterization: definition is inferentially ungrounded if and only if the procedure required to establish the applicability of the definiens consists in, or contains as a proper part a subprocedure that consists in, the very procedure required to establish the applicability of the definiendum.
To make the distinction vivid, Burgess offers the following two, admittedly unsatisfactory, definitions of cow: Footnote 23 continued the variables themselves. One would, for example, need stipulate that the variables themselves had an order. The same thing, though, would be possible with nominal predicates. 24 This might be too strong. The idea that definitions that are circular in this way are illegitimate is itself controversial. Gupta and Belnap (1993) develop a logic of circular definitions which they employ in analyzing the concept of truth. It would be interesting to explore this account of names in the context of a logic that allows circular definitions, but this would go well beyond the project here. Name-bearing, reference, and circularity 221 (8) (a) x is a cow = def Prince Charles knows that x is a cow. (b) x is a cow = def Prince Charles believes that x is a cow. (2008, p. 221) (8a) is inferentially ungrounded. The factivity of knowledge means that establishing the applicability of the definiens requires, in part, establishing the applicability of the definiendum.The non-factivity of belief means that (8b) is not inferentially ungrounded. One need not establish that x is a cow to establish that Prince Charles believes that it is. The nonfactivity of presupposition means that (6) is not inferentially ungrounded and so not viciously circular, at least according to Burgess' lights. Attempting to apply it will not launch a speaker on an inferential loop. The situation seems somewhat more complicated than Burgess makes it out to be. (8b) invokes Prince Charles' beliefs about what is and is not a cow. And even if we recognise that we need not settle whether some individual is a cow to establish whether Prince Charles believes that it is, we might still wonder about the sort of evidence we might have about whether Prince Charles believes that some individual is a cow. One thing we might try to do to decide whether Prince Charles believes that xis cow, would be to try to think our way into the Prince's epistemic position-what evidence is available to him? And this is positively bewildering. The question of whether x is a cow, for the Prince, resolves into the question of whether he believes x is a cow. But the most natural way for him of addressing the question of whether he believes it is cow, is simply to return to the question of whether it is a cow. 25 (6) avoids this particular regress worry, if it is one, as well. The question of whether some speaker is presupposing that P does not resolve into a question about whether the speaker believes P. Presupposing that P does not require a speaker to take any particular stance on whether P. (6) also distributes the responsibility across groups of speakers. To decide whether an individual bears Alfred a speaker should not focus on his own presuppositions, but rather on the presuppositions of groups of speakers.
But framing the issue in this way does bring out another potential worry-a practical analog to the epistemic troubles of Prince Charles. There is a kind of practical circularity in the way that Alfred is defined. Suppose a speaker asks herself: should I behave for the sake of communication as if a satisfies Alfred? Given the nature of referential communication, she should do it only if her conversational participants are likely to behave in the same way. But they are in same boat: wanting to behave that way only if everyone else is likely to behave that way. This is just the nature of coordination problems. In the typical case-that is, in the case of referential communication involving a non-nominal predicate-speakers solve this problem by anchoring their speech behaviour in their conception of the distribution of the property expressed by the predicate (and in their conception of their interlocutors' conception, etc.). Speakers are willing to act as if x satisfies mayor because they believe that x is a mayor, and that everyone else believes that, and so on. In the typical case of reference involving non-nominal predicates, the suitability of a predicate for reference to an individual is grounded in speakers' shared knowledge of the distribution of properties over potential referential targets. That is, in the typical case, the practical coordination embodied in the presupposition in a context that x satisfies P is achieved on the basis of descriptive coordination-all of the participants in the context have reason to believe that x satisfies P, or have reason to believe that the others believe that x satisfies P, etc.
But if Alfred is suitable for reference to x this suitability cannot ultimately be grounded in speakers' shared knowledge of the distribution of properties over potential referential targets. After all, if speakers know that x satisfies Alfred, what they know is that there is group of speakers who presuppose that he does. It might seem odd to explain the suitability of Alfred for reference to x in some particular context solely in terms of speakers' knowledge of the typical suitability of Alfred for reference to x. If we ask why speakers are willing to behave as if Alfred satisfies x, we find beliefs about how speakers are willing to behave for the sake of communication all the way down. Why would anyone be willing to behave that way in the first place?
So characterizing the meaning of Alfred in terms of speakers' presuppositions about the extension of Alfred sets up a kind of practical circularity. Speakers' behaviour with respect to Alfred cannot be ultimately anchored in descriptive coordination. It's important not to overstate the point here. Given some established practice of using Alfred to refer x, a speaker joining the practice can treat Alfred like any other predicate: simply tailoring his behaviour to his justified conception of the distribution of the property expressed by Alfred. But at the bottom of the practice lies a kind of arbitrariness: confronted with some group of targets about which about which there is no antecedent referential practice, the meaning of a nominal predicate gives speakers no guidance with respect to how to use names in reference. Any distribution they choose will be the right one, because in adopting it, they will make it the case that the extensions of the nominal predicates line up with it. 26 This is how the proposed account captures the response-dependent character of name-bearing.
[I]t is convenient to have in circulation[…] a tag, a designation, which does not depend for its referential or identifying force upon any particular[…] position or relation, which preserves the same referential force through its objects changes of position or relation and has the same referential force for communicators who know the object in different connections and for whom quite different descriptions would be uppermost. (Strawson 1974, p. 38) But the uniqueness and immense pragmatic convenience of proper names in our language lie precisely in the fact that they enable us to, refer publicly to objects without being forced to raise issues and come to agreement on what descriptive characteristics exactly constitute the identity of the object. They function not as descriptions, but as pegs on which to hang descriptions. Thus the looseness of the criteria for proper names is a necessary condition for isolating the referring function from the describing function of language. (Searle 1958, p. 172) Name-bearing, reference, and circularity 223 The upshot is this: a practice of using Alfred to refer to x cannot have its inception in speakers' true belief that x satisfies Alfred. This leaves two options for getting the practice going: false belief and bootstrapping. And these two cases correspond to reference transfer and baptism, respectively. To take reference-transfer first: Europeans falsely believed that there was a group of speakers who typically presupposed that the island was in the extension of Madagascar. Intending to join the practice, they began to presuppose the same thing. And thus came into existence a group of speakers who presupposed that the island satisfies Madagascar; and thus the island came to satisfy Madagascar (with some qualifications to be introduced below).
To return to baptism: I suggested that this account of name-bearing-invoking presuppositions of groups of speakers-can explain how individuals can bear a name in virtue of being baptised with it. The picture is simple: baptising x with Alfred is a public avowal to act as if x satisfies Alfred. We can ask: why do such avowals, under the right circumstances, bring it about that x satisfies Alfred? In this approach, this question becomes, why do such avowals, under the right circumstances, bring it bring it about that other speakers will presuppose the same thing?
As with any problem of coordination, such avowals can give a larger group a decisive reason to act in the same way. Imagine that you and I need to meet tomorrow, and have no other pressing concerns. If I send you a note saying that I'll be in the pub at noon, I can expect you to be there. Similarly if I say I'm going to act as if x satisfies Alfred, given your interest in grasping my referential intentions, you should take utterances of mine involving Alfred, under the right circumstances, as reflecting my intention to speak about x, and you should, given your interest in having me grasp your referential intentions, use Alfred if you wish to me to conclude that you intend to say something about x. Given their interest in achieving referential communication, each speaker has a reason to act as if x satisfies Alfred just in case every other speaker is likely to act the same way. An avowal, under the right circumstances, to act in that way by a single speaker can give a group of speakers' decisive reason to act in the same way (in the next section I'll try to say a little more about what the right circumstances are). 27 Footnote 26 continued Similar remarks can be found in Evans (1982, pp. 379-380) (I was made aware of Strawon's remarks in Jeshion (2009), where Jeshion develops a different picture of the characteristic communicative function of names). The reflexive account developed above does justice to the arbitrary connection between names and bearers in a way that the typical version of PV cannot, in virtue of their central appeal to naming practices. 27 This is obviously a little delicate. It doesn't seem incoherent to imagine a culture in which there is a practice of baptising a child with a name that is never meant to be used-perhaps names are sacred in some way. To the extent that I can think of such a practice as a form of baptism, it seems like a practice which is in tension with itself. It is a way of setting up a practice and at the same time forbidding speakers to participate in it. We should say, I think, that such a baptism does indeed create a disposition to use the name to refer to the baptised individual, but that, like all dispositions, it can be masked by interfering dispositions (in this case, the disposition associated with the taboo on using the name). It strikes me that we need both dispositions in place to understand the sense in which such a practice represents a kind of restraint or denial. For there have to be something restraining the use of the name, there must be some tendency towards its use, otherwise there would no restraint but merely an absence.
Of course naming practices, like baptism, go well beyond simple public avowals of referential intentions; various cultural accretions may attach themselves to the simple avowal. But this is not surprising. Various cultural accretions attach themselves to any conventional practice; this need not change the basic structure of mutual interest and dependence that constitute it.
To reiterate: there is a kind of circularity here, but it is a form of practical circularity that is characteristic of cooperative conventional behaviour. A given speaker is willing to act as if x satisfies Alfred because he knows that other speakers are typically willing to act in the same way-and this willingness is not based on some antecedent match between the descriptive condition associated with the predicate and speakers' conception of the properties of the individual. Nominal predicates are, as it were, empty vessels, waiting to be filled with the coordinated intentions of a group of speakers. With a normal (i.e. non-nominal) predicate P, speakers' referential behaviour with respect to P is grounded in their conception of the extension of P. With nominal predicates, the situation is reversed. The extension of a nominal predicate N is a function of speakers' referential behaviour with respect to N; the practical coordination involved in referential communication with a nominal predicate is not grounded on descriptive coordination. The inclusion of x in the extension of a nominal predicate N represents an achievement on the part of some group of speakers-they have bootstrapped their way into coordinated expectations about how each of them will behave with respect to x and N.

Groups of speakers
Alert readers will no doubt have noticed a hedge at the center of the account developed above. The account appeals to what is 'typically' presupposed in contexts involving groups of speakers. It would be preferable to eliminate this appeal to typicality in favour of an explicit characterization of the relevant features of the situation. It will be impossible to completely remove the hedge-the phenomenon is vague so we should not expect a perfectly precise characterization of it-but we should make sure to put the hedge in the place it belongs.
One of the goals of the account was to make sense of the fact that an individual can come to bear name in virtue of a mistake on the part of a group of speakers-the case of Madagascar was our example. We captured this by developing an account according to which the extension of a name was a function of speaker's attitudes about that extension. But there is a danger of going too far here. We do not want to suggest that speakers cannot be in error about who bears which name.
Consider the following case: A and B are introduced to Alfred at a party, but they both mishear the introduction and come to believe that Alfred is named 'Alvin'. At this point there is a group of speakers, namely the group consisting of only A and B, who are willing to behave as if Alfred satisfies Alvin. Does this mean that Alfred now bears the name 'Alvin' in addition to the name 'Alfred'? Clearly not. We need our account of name-bearing to allow for the possibility that a group of speakers might be mistaken about an individual's name.
Similar situations are possible with other conventional regularities. Imagine that C and D independently go on a vacation to a small town in Britain at the same time. Through some quirk of fate, both are ignorant of the fact that drivers drive on the left-hand side of the road in Britain. Through some even more bizarre quirk of fate, whenever they drive in the small village in which they are staying, they are the only cars on the road. They each drive on the right, coordinating their driving as they would if they were in driving in America. What is the status of this regularity? Is it a convention? It seems unnatural to say so. A and B's coordination is merely accidental. Given the beliefs and preferences which explain their behaviour, it is merely luck which allows them to coordinate their behaviour.
It will clearly not illuminate very much to insist that contexts in which A and B presuppose that Alfred satisfies Alvin are atypical, and thus do not make it the case that Alfred bears the name 'Alvin'. What we need to know is what precisely about the situation prevents their attitudes from changing the extension of Alvin. Why distinguishes their case from the case of Madagascar?
What we should focus on is the relation between a group of speakers, on one hand, and the interests and beliefs of its members which explain their coordinated activity, on the other. Referential communication, like driving, presents agents with a recurring coordination problem. Speakers find themselves in different contexts, with different interlocutors, trying to make reference to a range of different individuals. Each has an interest in promoting regularities in groups of speakers of which they are a part. If there are such regularities, and speakers know that there are, then in an arbitrary context involving speakers in the group, participants will be able to coordinate their referential behaviour. In any given context, successful coordination will be explained partly in terms of each speaker's belief that the context is of a certain type-namely one involving members of the group in which the regularity holds. A and B's referential coordination is accidental because among the beliefs that explains it is the false belief that speakers of some wider group use Alvin to refer to Alfred.
This last point needs elaboration. A and B have a range of beliefs about the contexts involving them. They falsely believe that speakers in a broader category use Alvin to refer to Alfred. They truly believe of each other that they use Alvin to refer to Alfred. Why is the regularity impugned by the false belief rather than legitimated by the true one? The answer lies in their own conception of what their up to. The false belief is what guides their referential behaviour. From their own point of view, a context in which the two of them are the only participants is not relevantly different from a context in which another party-goer is also present. They would presuppose the same thing about the relation between Alfred and Alvin in both contexts. And in the wider context this presupposition would not be shared.
If this is correct, we should be able to imagine a continuation of the story such that Alfred came to bear the name 'Alvin' in virtue of the referential practice of A and B. The more that we imagine that Aand B simply do not care how everyone else refers to Alfred (suppose for example, they live in isolation, only speaking with each other, disdainful of outside contact), the more it seems appropriate to say that Alfred is called 'Alvin'-that's their name for him. If they are only interested in coordinating their referential behaviour with each other, then the fact that they falsely believe that others refer to Alfred in the same way becomes irrelevant. This case is now structurally like the case of Madagascar. A practice which began as an error can make itself legitimate. The stain of the original mistake has been washed away and the situation is now an instance of the commonplace practice of a small group of speakers having their own name for a particular individual.
Incorporating this wrinkle into the picture, we are left with: (9) x satisfies N iff there is a group of speakers G such that, in contexts involving members of G, speakers presuppose that x satisfies N because they know that the participants in the context are members of G. 28,29 This is a natural elaboration of the original idea. It represents the fact that to bear a name it is not enough that a group of speakers have the right set of presuppositions. Those presuppositions must not be accidental. This still leaves a range of significant questions. What sort of conception must a speaker have of the group of speakers in which he is participating? Clearly speakers often do not have a very determinate conception of which group of speakers is the relevant one with respect to some particular name-using practice. For many speakers I imagine that the answer to the question-''What group of speakers do you take to be participants in the practice of using Jim to speak about Jim''-would be-''People who know his name''. But the way they operated with the name would reveal a substantive conception of the relevant group. Speakers would not expect people who had never been introduced to Jim, but only knew him as a regular at the gym, to act as if Jim satisfied Jim. Nor would they expect people who had been introduced to Jim, but only knew him from work where he went by 'James' to act as if he satisfied Jim. More generally, this is a place where contingent cultural and linguistic norms shape the practice of using names. What count as natural or typical groups of speakers with respect to the practice of using a name may depend on idiosyncrasies of a practice. In certain cultures, individuals may bear one name with respect to their family and a different one with respect to non-family members. In others, individuals may take on a different name in professional contexts. None of this should come as a surprise. When faced with a recurring coordination problem in which any particular instance of the problem involves only a subset of some larger group of potential participants, it is natural that different regularities should develop among different subgroups. Coordination is easier to achieve and maintain among smaller and more unified groups. Subgroups might develop naturally on the basis of historical precedent, or artificially on the basis of stipulation or agreement.
Questions remain about how determinate a speaker's conception of a group must be and how much variation across different participants in the same group should be tolerated. It seems inevitable that some vagueness will remain. This is a feature, it seems to me, of any conventional regularity, so not a challenge to the proposed account of name-bearing. 30 We can return one last time to baptism. The challenge was to explain how being baptised with N can bring it about that an individual bears N. We showed in the last section how, by thinking of baptism as a public avowal to act as if x satisfies N, we can make sense of how a baptism can give speakers who witness it a decisive reason to act in the same way. The question remains: why should speakers who have knowledge of a baptism typically count as a group of the right sort to make it the case that x bears N? The upshot of this section is that a group is of the right sort if members' beliefs about regularities within it guide their successful use of a name to refer to an individual. According to this picture, the group of speakers who have knowledge of a baptism will naturally form such a class; this is simply because, in the typical case, it will be common knowledge among them that they have witnessed the baptism, and common knowledge that each has reason to act in the same way that the others do. Under these conditions, witnessing one person vow to act in a certain way gives everyone a decisive reason to act in the same way in a future context which involves anyone else who witnessed the same thing.
6 Summing up I've tried to do a few things. First, I tried to show that proponents of PV, motivated by a flight from circularity, have offered a skewed picture of the nature of namebearing. An account of name-bearing must illuminate what baptism and referencetransfer have in common. I've developed an account which understands namebearing in terms of the presuppositions about the extension of a nominal predicate operative in groups of speakers. This account does contain a form of circularity-it represents the extension of a name as a function of speakers' presuppositions about the extension of that name. But there is nothing vicious in this circularity. 30 An issue that would need to be addressed in a full account is the way that sense/reference issues crop up here. I have been suppressing this complication for the course of the paper, and doing justice to it would require another paper in its own right. I can gesture here at how this sort of picture would handle the issues. Broadly, there are two ways to go here: One could try to introduce some notion like a guise of an individual-which was something like a stable, shared mode of presentation of an individual-and simply relativize the account above to guises. So an individual x would bear a name N just in case there is some guise for x such that in typical contexts speakers can use N to referentially communicate about x under that guise.
I prefer a slightly more complicated approach here. I'm not sure that there is any useful theoretical notion of a stable, shared mode of presentation (here I am loosely following the approach in Heck (1995Heck ( , 2012. Rather we need to relativize guises to contexts of utterance. So, given a context of utterance we can think of a guise simply as a class of mental files of the participants of the context (what determines the class is, of course, a tricky matter). The account of name-bearing would then go something like this: an individual bears a name N just in case in typical contexts there is some guise for x such that it is part of the common ground that x satisfies N under that guise.
Presuppositions about the extension of a predicate are not themselves a function of that extension. The circularity associated with the meaning of a nominal predicate means that a practice of reference involving it cannot be based on some antecedent descriptive match between an individual and the condition expressed by the predicate. Speakers must bootstrap their way into such a practice in virtue of either explicit agreement (baptism) or the mistaken belief that one already exists (reference-transfer).
These results are important for two reasons. Most concretely, they show that PV offers a coherent picture of the meaning of names. PV holds that names are predicates which express name-bearing properties. It claims that this sort of picture is supported by narrowly linguistic evidence. But it's always possible that apparent theoretical advantages at one level of explanation are the result of having shifted tensions or inconsistencies to another level of explanation. 31 In the case at hand it is possible, and has seemed likely to many, that the putative syntactic and semantic advantages of PV come at the expense of an ultimately incoherent metaphysics of name-bearing properties. I've tried to show here that PV can provide a coherent, and even plausible, picture of the nature of name-bearing. If proponents of PV are correct, and PV offers important theoretical advantages as a theory of names, the results here show that there are no metaphysical roadblocks to adopting it. Though proponents of PV have long claimed this, it turns out that they were right for the wrong reasons.
Secondly, I've provided a more detailed account of potential forms of circularity involved in the characterization of the meaning of an expression than is typical in the literature on names, and a more detailed account of when such forms are vicious. This is an important issue beyond PV. Claims about when accounts are, or are not, viciously circular are common in philosophy, but are also difficult to evaluate. In the case of PV, defenders of the view have been vague about what would count as vicious circularity (and Kripke's initial claim is, itself, not particularly detailed). I hope the discussion here may shed light in other areas where potentially fruitful philosophical approaches labour under the cloud of alleged vicious circularity. 32 31 This is a recurring worry about the relationship between semantics and metaphysics, going back at least as far as Russell's discussion of Meinong. Contemporary versions include the worry that certain elegant semantic treatments of modality are committed to an implausible ontology of possible worlds, and that certain elegant semantic treatments of action sentences are committed to an implausible ontology of events. Elsewhere in philosophy, structurally similar problems often relate epistemology and metaphysics (most famously, Benacerraf's problem). 32 One place where it would be fruitful to compare the discussion here would be the case of dispositional accounts of colour terms (e.g., x satisfies blue iff x looks blue to perceivers under normal conditions). Such accounts have long been dogged by accusations of vicious circularity. Recently, authors have mounted a defense of them, at least from the accusation of vicious circularity (Byrne and Hilbert 2011). It would be instructive to compare the two cases. Name-bearing, reference, and circularity 229