1 Introduction

In the last decade a new debate concerning the foundations of reference and semantics emerged (see, among others, Almog 2012; Bianchi 2012, 2015; Capuano 2012; Kaplan 2012; Wettstein 2012). It mainly focuses on how to interpret Donnellan’s seminal works and, in particular, on how it differs from Kripke’s influential contributions to so-called “direct reference”. The driving question is whether reference is mentally driven (viz. based on Donnellan’s notion of having in mind) or on Kripke’s notion of causal chain. Although this debate may concern the (historical) interpretation of Donnellan’s seminal papers, I will focus on this “new” reading/understanding of Donnellan and how, as it is presented by the aforementioned philosophers, it differs from Kripke’s picture. Thus, in focusing on the way a tokened name relates to its bearer I will present these two apparent competing stories (also) reflecting different attitudes concerning the foundations of semantics. To do so, I will discuss a Kripke-inspired picture and the way it differs from a Donnellan-inspired one, and show that there is a tension between the views that: (i) the token of a name refers to the object conventionally (causally) linked with the tokened name and (ii) the token of a name refers to the object the speaker has in mind. In short, there is a divergence on whether reference is socially or mentally driven.

I will end up suggesting that Korta and Perry’s (2011) critical referentialism/pragmatics and their name-notion network conception may help in clarifying this tension (and possibly to evade it). In so doing, I hope to provide some clues on how to deal with the tension as presented by this new understanding of Donnellan. I will not, though, discuss the merits of this “new” interpretation of Donnellan. For argument’s sake, I take it for granted. Besides, I cannot deal, in a short paper, with all the insights of the rich critical pragmatics framework as presented by Korta and Perry in their (2011) book. Yet, if I am right, it should emerge that the critical pragmatics framework gives us some new ways to deal with some emerging issues in the philosophy of language and mind. My aim is also to put into a historical perspective how Critical Pragmatics can handle some recent issues that emerged in the understanding of the direct referentialist movement as it came to the forum in the seventies.

2 The (Very Rough) Frege-Russell Pictures

Frege (1892) argued that a name refers via the mediation of a mode of presentation (a sense) of the referent. The referent of ‘Aristotle’, Aristotle, must satisfy the sense expressed by ‘Aristotle’, whatever the latter may be. Russell (1912), on the other hand, claimed that the only genuine names are the ones we use for objects we are acquainted with. Roughly, the demonstratives ‘this’ and ‘that’ are genuine names when used to refer to the objects we are directly acquainted with. For simplicity’s sake, let us assume that the objects we are acquainted with are the ones we are presently perceiving. Thus, I am acquainted with the computer in front of me because I perceive it and I am acquainted with Mary because I see her. I can thus refer to the computer using ‘this’ and to Mary using ‘she’ or ‘that woman’. According to Russell we are acquainted with an object only when the mind enters into a direct unmediated relation with the thing perceived.Footnote 1 To be direct, this perceptual relation, the acquaintance relation, must be unmediated by a conceptual or descriptive intermediary. Since we are not acquainted with Aristotle, our use of ‘Aristotle’ cannot count as a genuine name. Ordinary proper names are disguised definite descriptions, Russell told us. As such, given Russell’s (1905) theory of descriptions, they cannot be tools of singular reference because, on the proposed view, ordinary proper names do not contribute their referent to the proposition expressed. Instead, as disguised descriptions, names contribute conditions the referent must satisfy. The proposition expressed is general, not singular.

The Frege-Russell divergence on reference-fixing is, as far as I know, the first time in the history of analytic philosophy that the social vs. the mental conception of reference emerged.Footnote 2 According to Frege, senses (the determiners of reference) are mind independent and must be grasped by people competent with the language they exploit when singling out objects of discourse. And language is social: it is something that the members of the linguistic community share. On the other hand, for Russell singular reference is epistemologically driven and rests on a strict epistemic notion of acquaintance.

Russell’s conception has somewhat been equated with Frege’s: Frege suggests that the sense of a proper name can be expressed by a (or some, a cluster of) definite description(s).Footnote 3 The Frege-Russell theory has often been labeled (see, e.g., Donnellan 1970; Kripke 1972) a descriptive theory of reference. On such a view the referent of a given name must satisfy the description(s) it expresses as a sense (Frege) or the description replacing the proper name (Russell). It is not my intention to enter into the merits and demerits of Frege and Russell’s respective theories. Nor is it my intention to discuss the vast criticisms such a theory encountered in the seventies (see, in particular, Donnellan 1966, 1970; Barcan Marcus 1986; Kripke 1972; Kaplan 1977; Perry 1977; Putnam 1975). It suffices to mention that Kripke and Donnellan cogently argued that we could effectively refer to individuals with names even if we do not express uniquely identifying description(s) of the individuals in using those names. In using ‘Thales’, I can refer to Thales and talk about him even if I know nothing or close to nothing of him. Even if I do not know whether Thales was a man, a woman or a building. I can conjecture whether Aristotle had a tattoo on his left arm and say “Aristotle had a tattoo on his left arm” and in so doing refer to, talk about, and think of Aristotle. I can also say: “Aristotle could have died in childhood and if he did he would never have been a pupil of Plato”. Yet, if ‘Aristotle’ refers via, or is replaced by, ‘the pupil of Plato’, I would express a contradiction amounting to saying that it could have been that the pupil of Plato who tutored Alexander was not a pupil of Plato.Footnote 4 In his critique of Russell’s and Quine’s view on proper names, Geach anticipated much of this discussion:

[W]hen I refer to a person by a proper name, I need not either think of him explicitly in a form expressible by a definite description, or even be prepared to supply such a description on demand (not, that is, with any confidence that the description really is exclusive). (Geach 1957, pp 67–68)

3 The Causal Chain Picture

The descriptivist conception of reference has been ousted by the so-called “causal theory of reference” or “causal chain of reference”. The aims of such a theory (or picture) are to explain: (i) how a term acquires its specific referent and (ii) how it is connected to its bearer. In the case of proper names the causal picture usually runs as follows: a name’s referent is fixed by an original act of dubbing or tagging (by a sort of initial baptism) and subsequent tokens of the name succeed in referring back to the referent by being linked to that original dubbing via a causal, historical, chain. It can be that the causal chain ends, to use Donnellan’s (1974) terminology, in a block. This would be, for instance, the case when a fictional name is introduced (e.g. when Conan Doyle introduced the name ‘Sherlock Holmes’) or some entity is stipulated (e.g. when Urbain Le Verrier stipulated the existence of Vulcan as a planet disturbing Mercury’s orbit), i.e., when the dubbing fails to name an individual. A name may come into existence in various ways. As Donnellan puts it: “the first use of a name to refer to some particular individual may be in an assertion about him, rather than any ceremony of giving the individual that name” (Donnellan 1974, p. 113, footnote 13).

To the best of my knowledge the first philosopher who proposed the causal theory of reference (or causal chain) is, once again, Geach:Footnote 5

I do indeed think that for the use of a word as a proper name there must in the first instance be someone acquainted with the object named. But language is an institution, a tradition; and the use of a given name for a given object, like other features of language, can be handed on from one generation to another; the acquaintance required for the use of a proper name may be mediate, not immediate. Plato knew Socrates, and Aristotle knew Plato, and Theophrastus knew Aristotle, and so on in apostolic succession down to our own times; that is why we can legitimately use ‘Socrates’ as a name the way we do. It is not our knowledge of this chain that validates our use, but the existence of such a chain; just as according to Catholic doctrine a man is a true bishop if there is in fact a chain of consecrations going back to the Apostles, not if we know that there is. (Geach 1969/1972, p. 155)

Kripke defends a similar view that, to give it a name, can be characterized as a social conception concerning the way names relate to their bearers.Footnote 6 We refer to Aristotle using ‘Aristotle’ because of previous uses of the name. The name belongs to the common language and, to borrow Kaplan’s (1989) terminology, it comes prepackaged with its semantic value. A name is historically related to its bearer and our tokening of it inherits the name’s semantic value from previous uses. In short, we refer to (and think of) Aristotle vicariously:

In general our reference depends not just on what we think ourselves, but on other people in the community, the history of how the name reached one, and things like that. It is by following such a history that one gets to the reference. (Kripke 1972/1980, p. 95)Footnote 7

On this view a name’s reference is borrowed from previous uses. Any token of ‘Aristotle’ brings us back to Aristotle because he has been so-dubbed. This does not amount to say that ‘Aristotle’ refers to the performative act of dubbing or baptism. The latter is just the historical fact that introduced the name for Aristotle into the linguistic community, what initiated the practice of using ‘Aristotle’ for Aristotle. What matters, for reference, is the existence of a permissive convention or link that allows speakers in the linguistic community to exploit it. In other words, what matters is the existence of a practice sustaining the use of the name that speakers participate in.Footnote 8 The way this practice came into existence is not semantically relevant. It belongs, to use Kaplan’s (1989) terminology, to meta-semantics.Footnote 9 Think, for instance, of name changes in the course of someone’s life. After marrying John Taylor, Mary Smith became Mary Taylor. She did not change her identity; ‘Mary Smith’ refers to the same individual as ‘Mary Taylor’. All she did was to acquire another (non-substantial) property, i.e. another name. A new convention is thus created. We can now say that when six years old, Mary Taylor was a good student. With ‘Mary Taylor’ we can thus refer to Mary Smith independently of her changing the name. As Geach forcefully puts it:

[T]he proper noun ‘Augustus’ as used in Roman history books has Octavian for its bearer; this is true without temporal qualifications, even though Octavian lived for years before being called by that name; it would be absurd to object to the question ‘When was Augustus born?’ because the name was not conferred on him then. (Geach 1962, p. 29)

Dummett (1973/1981, p. 183) makes the same point: “proper names are temporally as well as modally rigid”. The temporal rigidity of proper names is even more evident if we consider geographical names. We can, for instance, say: “Two thousand years ago Cuba was inhabited by aboriginal people, while nowadays it is mainly occupied by people of African and European origins”, even if the name ‘Cuba’ came into existence only after the arrival of Columbus in 1492.

Furthermore, through history the name may have changed in its writing and/or pronunciation. Yet it is the same name-using practice that brings us back to Aristotle. As semantics is concerned we face here what Kaplan (1989) characterizes as consumerist semantics. For, we are, most of the time, language consumers. Only in a dubbing episode are we language creators. When Aristotle’s parents dubbed their child ‘Aristotle’ (they probably used ‘Αριστοτέλης’ or something similar) they initiated a convention, a network allowing us to refer back to their child using ‘Aristotle’. It is in virtue of this convention that we can now think and talk about Aristotle. So goes the social conception picture of reference, as I understand it.

4 The Donnellian Picture

Donnellan favors another picture that, following Kaplan (1989), can be characterized as subjectivist semantics. For his “historical” explanation seems to leave no room for reference borrowing, i.e. the view that my token of a name inherits its reference from previous tokens of that name. Each time we token a name we seem to fix its reference anew. For, if reference is driven by the having in mind, then it is the object the speaker has in mind when she tokens an expression that ends up as a constituent of the proposition expressed (see Almog 2012; Bianchi and Bonanini 2014).Footnote 10 This line of thought is substantiated by Donnellan’s assimilation of proper names to his treatment of the referential use of definite descriptions: “my account of proper names … seems to me to make what I called ‘referential’ definite descriptions … a close relative of proper names” (Donnellan 1970, p. 78, note 8).

When we use a definite description referentially, the individual we refer to is the one we have in mind. And our reference-fixing can be successful even when the relevant individual, the referent, does not satisfy the descriptive content of the description voiced to single it out. In such a case the description is just a tool used to identify the object. The speaker may have used any other expression to perform the same job.Footnote 11 Donnellan’s paradigmatic examples illustrating the referential use of descriptions are made using misguided descriptions—e.g.: the speaker successfully referring to a man drinking water using ‘the man with the Martini’.Footnote 12 One can successfully pick out the relevant man even if she knows that he is not drinking a Martini:

It is also possible to think of cases in which the speaker does not believe that what he means to refer to by using a definite description fits the description, or to imagine cases in which the definite description is used referentially even though the speaker believes nothing fits the description. (Donnellan 1966, pp 13–14)

This choice is revealing on how Donnellan takes the having in mind as the key notion and the starting point of a theory of language and communication. With reference via singular terms, the primary function of language is to single out objects of thought and in a communicative interaction to pass to the audience the objects the speaker has in mind. Furthermore, in characterizing the referential use of definite descriptions, Donnellan appeals to Russell’s (1912) notion of acquaintance and assimilates the referential use of descriptions to Russell’s genuine names:

[O]n Russell’s view the type of expression that comes closest to performing the function of the referential use of definite descriptions turns out, as one may suspect, to be a proper name (in “the narrow logical sense”). Many of the things said about proper names by Russell can, I think, be said about the referential use of definite descriptions without straining senses unduly. Thus the gulf Russell thought he saw between names and definite descriptions is narrower than he thought. (Donnellan 1966, p. 4)

Genuine proper names, in Russell’s sense, would refer to something without ascribing any property to it … when a description is used referentially, a speaker can be reported as having said something of something … we are concerned with the thing itself and not just the thing under a certain description, when we report the linguistic act of a speaker using a definite description referentially. That is, such a definite description comes closer to performing the function of Russell’s proper names than certainly he supposed. (Donnellan 1966, p. 27)

Like Russell, Donnellan assumes that acquaintance is both what helps us to entertain singular thoughts and what undermines Frege’s notion of senses. Like Russell, Donnellan subscribes to the doctrine of direct realism, viz. the view that the mind can enter into direct contact with the external world.Footnote 13 We can thus have direct and unmediated knowledge of objects.

How does the mind, with the use of proper names, enter into a direct contact with the external world, i.e. with the name’s referent? To begin with, we should stress that it is the token of a proper name that refers. Names, in themselves, are inert.Footnote 14 The object/referent of a description used referentially is the object the speaker is acquainted with, the object that the speaker has in mind, the object that directly reaches the cognizer. Thus, if a tokened proper name is similar to a tokened description (used referentially), a speaker refers to the object she has in mind and the speaker’s thought is object-dependent or de re:

[I]f one says, for example, “Socrates is snub-nosed”, the natural view seems to me to be that the singular expression ‘Socrates’ is simply a device used by the speaker to pick out what he wants to talk about while the rest of the sentence expresses what property he wishes to attribute to that individual. (Donnellan 1974, p. 90; my italics)

At this point one could argue, inspired by Kaplan’s (1989) consumerist semantics and Kripke’s (1980) causal chain, that one has Aristotle in mind insofar as one tokens the name ‘Aristotle’ and, in so doing, inherits its semantic value. Names come to us prepackaged with their semantic value, Kaplan told us. This, though, does not seem to square with Donnellan’s narrative concerning the referential use of definite descriptions. As we saw, such a use is equated to Russell’s genuine names, whose reference is fixed via the acquaintance relation. The reference relation is direct insofar as the object itself impinges on the speaker’s mind. If we start with Donnellan’s notion of having in mind as the basic building block of reference-fixing, then it seems that for Donnellan to accept a Kripke-like picture concerning the causal chain, one ought to argue that a use of ‘Aristotle’ makes one acquainted with the Greek philosopher. This is a plausible move (see e.g. Recanati 2010). In other words, the notion of acquaintance must be severed from Russell’s strict epistemic requirements. In short, one’s thought can be about an individual insofar as one is engaged in a practice to use a given name that stands for that individual. One’s singular thought can thus be parasitic on the common language one is using. Yet, it is the object one has in mind that triggers one to utter a given name.

Before going further it is worth mentioning how Kripke’s causal chain story primarily focuses and rests on the deferential use of proper names and, as such, seems to insist on reference borrowing. Kripke’s many examples mention uses of names of famous (and some infamous) persons: ‘Hitler’, ‘Feynman’, ‘Einstein’, ‘Gödel’, etc. Although such uses are quite common, they do not constitute our everyday paradigmatic use of proper names. The latter concerns primarily names for people (or objects) we are familiar with. That is, to use Donnellan’s idea, of people we have in mind (and are acquainted with). Many proper names we have in our idiolects are names of friends, family members, colleagues, pets, places, etc. Such uses concern individuals we know quite a great deal about, that is, individuals we can easily pick out and/or identify among the many we are surrounded by. We do not often use names we casually pick up in supermarkets or by looking into the phone book. When we hear a name that does not bring some interest or salience to us we do not store it in our long-term memory. One may know close to nothing about Aristotle or Feynman. Yet one knows quite a lot about one’s sibling, partner, colleague, daughter or town.Footnote 15 This does not amount to saying that the reference of a given name is fixed by the descriptions (or information) one can associate with it. Reference by singular terms is “unmediated”. It is not determined, as Frege supposed, satisfactionally. That is, the referent of a name is not the individual that happens to satisfy the sense expressed. It means, though, that the notion of having in mind plays a crucial role in a speaker’s token of a proper name. It is the token of the name that refers and carries both the object the speaker has in mind and the individual that, in the context of communicative exchanges and thought episodes, enters the proposition expressed, roughly, what is said or Kaplanian content. The object that enters the proposition expressed is not the one that satisfies the descriptive content the speaker may supply. It is, rather, the one that lays at the beginning of the causal or historical chain, i.e. the object that, thanks to the name entering the convention exploited in the linguistic community, links the having in mind to the bearer. The historical chain is, to use Donnellan’s terminology, a causal chain, which runs independently from the psychological vagaries of the speaker-hearer. In a causal chain only tokened names can be causally related to their bearers. That is, only uttered or thought words qua episodes or worldly events can enter a causal, physical relation. Following this line of thought, we could say that the speaker is acquainted with, say, Aristotle, insofar as she comes to have Aristotle in mind. The way one comes to have an individual in mind, though, may not rest on a direct perceptual encounter with him. One may form a vivid notion of Aristotle in her mind through various channels, e.g. by reading his books, his biography, by attending classes on him, etc. In such a case her token of ‘Aristotle’ makes her acquainted with the Greek philosopher inasmuch as she possesses a notion (a file) of Aristotle in her mind. This file relates to Aristotle in a causal, relational, way. And the speaker may thus entertain a de re thought about Aristotle. The thought is causally related (via the historical-causal chain) to Aristotle.Footnote 16

5 Speaker’s Reference or Semantic Reference?

To further highlight the difference between Kripke’s and Donnellan’s accounts we can appeal to a renowned example given by Barwise and Etchemendy (1987, p. 29). If John utters, describing a particular poker hand he is looking at, “Claire has the ace of hearts”, he picks out the relevant player he has in mind, the one he intends to talk about, even if the latter is not Claire.Footnote 17 For sure, something went wrong insofar as John misidentifies the relevant player with Claire. Yet, to correct John the hearer (in the known) must first think and pick out the relevant player and only in a second time argue that John mistook the poker-player to be Claire. Furthermore, for John to be able to retrieve his use of ‘Claire’ after realizing he made a mistake he must go back to the object he received information from, i.e. the relevant poker-player he mistook to be Claire (the object he had in mind when he voiced ‘Claire’). What John ends up saying may be false (if the relevant player does not have the ace of hearts). And it is false even if it happens that Claire is not playing in that game and she is playing poker in another town with the ace of hearts in her hand. Claire is out of the relevant thought episode and communicative situation. This explanation fits Donnellan’s account: John refers and talks about the player he has in mind. John’s token of ‘Claire’ is causally (in our example via direct perception) related to the individual he focuses his attention toward, to the individual that directly reaches his mind and he is gaining information from.Footnote 18 If one were to buy into Kripke’s causal theory what John says is true insofar as he, inheriting the semantic value from previous uses of ‘Claire’, refers to Claire, who happens to have the ace of hearts. It would be hard to claim that in such a situation John does not have in mind the player he is perceiving and gaining information from. As the layperson’s intuitions go, John has in his mind the player he is looking at and is talking about and John says of this relevant player that she has the ace of hearts. In that case ‘Claire’ works like a description used referentially and it can pick out the relevant individual even if it does not satisfy the property of being called ‘Claire’ (or carrying the name ‘Claire’): ‘Claire’ works like a Russellian genuine name. As such, ‘Claire’ is just a tool used by John to individuate the relevant poker-player, the object he has in mind and intends to talk about. For John’s communicative plan is to individuate that player and attribute to her the property of having the ace of hearts. Instead of using ‘Claire’ he could have uttered ‘that player’ or a description used referentially such as ‘the player with the blue shirt’. In such a case, if the descriptions or demonstrative expression fit the referent we would have no problem so that John would not retrieve the label he used in his referential act when realizing that he made a mistake. The reference and the object John has in mind is determined relationally (in this example through direct perception). Thus, John entertains a singular, de re, thought about the relevant individual he is gaining information from. It is the causal (perceptual in the example) relation that makes the relevant player the object John has in mind and the referent of his token of ‘Claire’.Footnote 19

One way to defend Kripke’s causal chain and to settle the debate would be to appeal to his (1977) famous distinction between semantic reference and speaker’s reference. Thus, while John, the speaker, refers (speaker’s reference) to the relevant player, viz. to the one he intends to talk about, the semantic reference is Claire and what he says is, from a semantic viewpoint, true.

Kripke’s notion of speaker’s reference is alien to Donnellan’s overall program. First, Donnellan never framed his distinction between the referential and the attributive use of definite descriptions within a framework appealing to speaker’s vs. semantic reference. As Bianchi and Bonanini (2014) point out, there is no evidence in Donnellan’s work that he considered his reflections on proper names (and the referential use of descriptions) as not being semantically relevant.Footnote 20 Quite the opposite: Donnellan’s claims about the referential use of definite descriptions (not to mention indexicals) and on proper names are semantic in nature. Secondly, Donnellan’s aim is to stress how speaker’s reference and semantic reference cannot be severed. This last claim is further highlighted by Donnellan’s (1978) reflections on anaphora. Kripke’s (1977) critique of Donnellan’s distinction and, in particular, Kripke’s charge that the referential use does not belong to semantics but is a pragmatic fact, misses Donnellan’s central point. For the referential use is considered to be of semantic significance. In considering anaphoric pronouns linked with descriptions, Donnellan argues “that speaker reference cannot be divorced from semantic reference” (Donnellan 1978, p. 116). Geach anticipated this:

My use of ‘he’ in this case [‘There is a man in the quarry-edge’; five minutes later I say ‘Now he’s gone—he must have fallen in!’] has nothing to do with conversio ad phantasmata; ‘he’ does not get its meaning from anything that could be sensibly indicated. Crudely: ‘he’ does not mean ‘the man I am now looking at, or pointing to’ but ‘the man I meant a little while ago’; what is required in my referring to him as ‘he’ is not a present sense-perception but a recent thought. (To use Aquinas’s language; the pointing, demonstratio, effected by the demonstrative word ‘he’ is not ad sensum but ad intellectum). (Geach 1957, p. 74)

Donnellan’s (1978) treatment of anaphoric pronouns linked to a description, like ‘she’ in: “One of the philosophers I saw last week came to my office today. She gave me a box of chocolate”, points toward this. The description is improper since it does not single out a specific individual. Yet, for the anaphoric ‘she’ to pick up an individual it must be linked with the individual the speaker has in mind. Yet, if the description is (semantically) explained, after Russell and Kripke, in quantificational terms, ‘she’ does not pick up an individual. It is the having in mind that does the trick:

[I]n these examples some particular person or persons are being talked about and the definite descriptions and pronouns seem surely to have particular semantic referents. If the descriptive content of the uttered descriptions even augmented by background assumptions, etc., are insufficient to determine the referents, how is this possible? My answer will not be unexpected. The speaker having some person or persons in mind to talk about can provide the needed definiteness. Once more, then we have a series of instances in which speaker reference appears necessary to provide semantic reference. Hence, not just to provide the right reference, but to allow reference at all. (Donnellan 1978, pp 134–135)

The question I now turn to is whether there is a unique account that represents our cognitive make-up and our linguistic interchanges in using proper names.

6 Names and Having in Mind

The notion of having in mind plays an important (semantic) role and it constitutes the starting point, the main building block—Capuano (2012) claims that it constitutes the ground zero of semantics—for a theory of direct reference (see also Almog 2012). Furthermore, as Kaplan (2012) puts it, Donnellan’s notion of having in mind does not rest on one having a proper name in her idiolect. It does not necessarily rest on one perceiving (or having perceived) the referent. Someone can have an object in mind in many different ways. The having in mind rests on one having a notion (or a file) of the referent in one’s mind. Proper names, like any other device of so-called singular (direct) reference, be it reference by indexicals, descriptions (used referentially), etc., point toward the view that the alleged speaker’s reference trumps so-called semantic reference and that it is the former, if we were to buy into the speaker’s/semantic reference distinction, that is the basis of a theory of reference, thinking and communication:

Donnellan once said to me that he could imagine the name ‘Aristotle’ having been first introduced in the Middle Ages by scholars who previously had used only definite descriptions to write and speak about Aristotle. According to Donnellan, these scholars may well have had Aristotle in mind, and through their conversations, through the referential use of definite descriptions and other devices, passed the epistemic state of having Aristotle in mind from one to another. Thus they were properly situated from an epistemic point of view to be able to introduce a proper name. (Kaplan 2012, p. 142)

Surprisingly, in his latest (2012) publication Kaplan has changed his view. He now seems to disavow both a Kripke-like causal chain of reference, as well as his (1989) consumerist semantics. Names are helpful tools to transmit what we have in mind, yet they are inessential. Having a name may be a sufficient condition for having someone or something in mind; it is not, though, a necessary condition. How often one says “I’ve got her name on the tip of my tongue” intending to pass to one’s audience the individual one has in mind? Or one says: “I forgot her name though I know a lot about her” or “I see whom you mean but I do not know her name”. In such cases the speaker has someone in mind and intends to talk about her without being capable of uttering her name. Having someone or something in mind may thus be independent of having their name in one’s idiolect. Though language is what permits us to talk and think about past individuals, the presence of proper names may play no (essential) role in our ability to transmit a having in mind. Names may be a sufficient condition for having someone or something in mind; they are not, though, a necessary condition. After all, many things we have in mind bear no name. Kaplan recognizes that in his previous works he “had the relation between names and having in mind backward. The name rides on the having in mind, not the reverse” (Kaplan 2012, p. 149). When we come to describe Donnellan’s referential use of descriptions, the having in mind is prior to the uttering of the description. The description, one could say, is just a (inessential) means to pass to our audience the object we have in mind.

If the use of a name rides on the having in mind, we can then ask: how does the token of a name relate to its bearer?

We seem to face a dilemma. On the one hand, it appears that a token of ‘Aristotle’ relates to Aristotle inasmuch as the speaker has Aristotle in mind. But what does it mean to have Aristotle in mind? One can have Aristotle in mind in many different ways. In Donnellan’s scenario, before introducing the name, the medieval scholars had Aristotle in mind and were able to think and pass along information about him without using ‘Aristotle’. They were passing along information concerning Aristotle into a network of information. On the other hand, it seems that we have Aristotle in mind because we token ‘Aristotle’ which, through a reference chain or network, relates to its bearer. We seem to face a tension (yet, as we will see in the next section this tension can be dealt with the adoption of Korta & Perry’s critical referentialism/pragmatics). For in some cases a name rides on the having in mind, while in other cases the having in mind rides on the name. As we will soon see, a plausible theory of communication must accommodate both views, i.e. that names ride on the having in mind and that the having in mind rides on names.

People can have someone in mind insofar as they have a notion (or file) of the relevant individual in their mind.Footnote 21 And the relevant notion or file need not necessarily carry a name. We certainly have many people and things in mind without knowing their name; we may recognize them, we may know quite a lot about them (e.g. what they look like, where they work, recognize their voice, have an image of them in our mind, etc.) and can pass on the individual we have in mind using various kinds of descriptions or demonstrative expressions such as, e.g., ‘the funny bartender we met last night’, ‘that old man we just saw behind the counter’, ‘the woman with the blue hat you talked with this morning’, etc. In short, the conveying of information about individuals need not be under a given name inasmuch as there are different ways we can think of and identify them: we may have description-based, demonstrative-based, information-based, etc. ways to think and talk about them. Yet, in many cases we gather information about individuals under a given label. In such instances proper names are particularly useful. Proper names also allow us to have someone in mind in a deferential way. Once again, claiming that one refers to someone because one has the relevant individual in mind does not amount to saying that the reference is fixed by the information one had in her mind; reference is fixed relationally, not satisfactionally.Footnote 22 I can have Aristotle in mind because I token ‘Aristotle’ and, in so doing, I refer to Aristotle, because ‘Aristotle’ connects via a network or chain of reference to Aristotle. I can refer to my five-month still-born child because I gave her a name even if the name has never been voiced in the public sphere and never entered a social practice. Yet, had I voiced the name and introduced it to my family or friends, others could refer to and think of her using the name. I can also think and entertain singular thoughts about her even if I never gave her a name. Being amnesiac I can think about and refer to myself because I have my name in mind even if I do not know that in voicing it I refer to myself and I do not know that the information I associate with this name concerns me, and so on and so forth.

7 The Name-Notion Network

The notion of having in mind takes center stage when dealing with problems pertaining to cognitive significance, communication, explanation of behavior, etc. Thus, a theory of meaning aiming to account for communication and understanding must be, to borrow Korta and Perry’s (2011) terminology, utterance-bound and ought to consider the variegated ways a subject can have someone/something in mind. Besides, there can be different ways one can have the very same object in mind even in the case of direct perception (acquaintance-based) relation to it (see e.g. Kripke’s 1979 well-known Peter-Paderewski case). The having in mind is, particularly in the case of perspectivally driven perception, what anchors one’s thought to the external world. Since the very same object can be apprehended (and thus represented) in different ways the referent can affect the cognizer’s mind in variegated ways. And it is how the agent comes to have an object in mind that helps in dealing with puzzles pertaining to cognitive significance. How someone comes to have an object in mind is also what helps in explaining mental causation and, thus, someone’s behavior. To do so, we can focus on different contents associated with a given utterance when we come to characterize and classify what goes on in an agent’s mind. It is at this point that Perry’s critical referentialism and Korta and Perry’s critical pragmatics becomes relevant. A simple utterance comes equipped with different contents (propositions or truth-conditions). We can characterize different aspects and properties of an utterance. On the one hand, we can take episodes of speech or thought themselves, such as the time and the place of the episode, as content constituents. In so doing we can classify the representations involved. On the other hand, we can put conditions on the subject matter, the proposition expressed (or Kaplanian content). As Kaplan puts it: “We use the manner of presentation, the character, to individuate psychological states, in explaining and predicting action” (Kaplan 1989, p. 532; see also Perry 2001/2012).

I would now like to highlight how critical referentialism (or critical pragmatics to use Korta and Perry’s terminology) relates to the notion of transmission of information and to what Perry (2001/2012) and Korta and Perry (2011) characterize as the name-notion network. As we saw, Donnellan and Kripke hold different views concerning the so-called causal chain of reference. While Kripke focuses on the way a tokened proper name inherits its semantic value from previous uses, Donnellan (at least following the new interpretation highlighted by Kaplan 2012; Almog 2012; Bianchi 2012, etc.) seems to suggest that each time we token a name we fix the reference anew. The question I would like to raise is how the name-notion network fits within these two alternative conceptions. On the one side, Korta and Perry seem to subscribe to the Kripkean view when he claims that his use of ‘Aristotle’ is linked to the Greek philosopher because, through a network of communication, it is related to the referent. On the other side, Korta and Perry insist about information transmission and how we come to have notions stored in our mind (see Perry’s conception of mental file).Footnote 23 This amounts to saying, in the critical referentialism parlance, that the official (referential or Kaplanian) content and the reflexive content(s) triggered by the utterance (the speaking episode) are both crucially important when we classify what a speaker said and how she said it. The what and the how cannot be split (on the referential/reflexive content distinction see Perry 2001/2012, more on this in the next section).

In assuming critical referentialism or critical pragmatics the Donnellan-Kripke debate can now be apprehended under a different light. The Perry and Korta and Perry-inspired view can be understood as a sort of hybrid of the two positions. For the name-network and the notion-network need not be mutually exclusive; they work in tandem and tend to proceed hand in hand. Depending on the speaker/thinker’s cognitive situation and her intentions, sometimes it is the notion-network that guides the referential link, while other times it is the name-network. Kripke’s causal chain pertains to what Perry and Korta and Perry characterize as the official or referential content (roughly what is said) and, in normal circumstances, determines the object the speaker has in mind and enters the proposition expressed, the referential content. This is the name-network. On the other hand, the notion-network is what delivers how the speaker gets the relevant referent in mind. Furthermore, it is the notion-network that allows a speaker to cumulate and pass on information concerning a given referent. Most of the time we face no problems. The object one has in mind is the object that, through the name-network, is referred to. It can happen, though, that the name-network and the notion-network break apart. In some cases, as in Donnellan’s example concerning the medieval scholars introducing the name ‘Aristotle’, people can pass on information (and the individual they have in mind) without having the name in their idiolects. Thus, the passing of information from one speaker to the other, the passing of whom they have in mind, can be sustained by a notion-network without having to appeal to a name-network. The name-network and the notion-network, in some awkward cases, can run in different directions (as in, e.g., Barwise and Etchemendy’s example of Claire and the ace of hearts).Footnote 24

As I suggested, Kripke endorses what Kaplan characterized as consumerist semantics, while Donnellan subscribes to a form of subjectivist semantics. Within Korta and Perry’s critical referentialism/pragmatics framework, as I understand it, speakers are both consumerists and subjectivists. They are consumerists for they are embedded in name-networks, and they are subjectivists inasmuch as they are entrenched in notion-networks. In other words, they are consumerists inasmuch as they refer to, say Aristotle, by using the public name ‘Aristotle’, while they are subjectivists insofar as they have a notion (or file) of Aristotle in their mind and are thus embedded in a notion-network.

One is compelled to choose between a Kripke-style or a Donnellan-style chain of reference if one embraces mono-propositionalism, i.e. the view that all the relevant information pertaining to a communicative or thinking episode is encapsulated into a (unique) content or proposition expressed.Footnote 25 Thus, we have to choose whether it is the proposition the speaker semantically expresses or the one that she intends to communicate (having the intended referent as a constituent) that contributes in the passing of information from one speaker to her audience. If one embraces critical referentialism/pragmatics one need not choose between these two horns. The proposition expressed by a given utterance differs from the one “created” (see Perry 1988) or the reflexive truth-conditions/content (see Perry 2001/2012; Korta and Perry 2011). Both propositions can contribute in the passing of information and in the success of communication. While the reflexive contents focus on the notion-network, the referential (or official content) concerns the name-network. Thus, Kripke-style and Donnellan-style chain of reference and communication can sit side by side when we come to explain communication, problems pertaining to cognitive significance, reference-fixing and information-transmission. The picture, though, is more complicated than it first appears.

8 Causal Chains and Having in Mind

Let us consider, once again, Barwise and Etchemendy’s scenario where, using ‘Claire’, John refers to the poker-player he has in mind, i.e. the one he perceives and intends to talk about. In such a case ‘Claire’ works as a Russellian genuine name (or as a description used referentially). If we stick with Perry’s original distinction between reflexive truth-conditions and official truth-conditions we would have the following analysis:

  1. (1)

    Claire has the ace of hearts

The referential or official content of this utterance would be the proposition that Claire has the ace of hearts, while the reflexive content is:

  1. (B)

    There is an individual x and a convention C such that:

    1. (i)

      C is exploited by (1)

    2. (ii)

      C permits one to designate x with ‘Claire’

    3. (iii)

      x has the ace of hearts

If this were the case, though, John, exploiting the convention linking ‘Claire’ with Claire, would refer to Claire who is out of the communicative situation and is holding the ace of hearts in another poker game. Yet, if we follow Donnellan’s insight, John is not referring to Claire but to the poker-player he has in mind. Thus, for ‘Claire’ to refer to the relevant poker-player John has in mind the link between ‘Claire’ and the referent is not, and cannot be, secured by convention C. It is secured by the having in mind. In such a situation the having in mind trumps convention C.

The relevant question we now face is: how can the having in mind be represented at the reflexive level? The following representation should do the job:

  1. (C)

    There is an individual x the speaker of (1) has in mind (and intends to talk about) such that:

    1. (i)

      the speaker of (1) utters ‘Claire’ to designate x

    2. (ii)

      x has the ace of hearts

While (2) captures the name-network, (3) should represent the notion-network. While (2) focuses on the reflexive content of the utterance, (3) focuses on the reflexive content of the thought triggering the speaker’s utterance of (1). In our example (3) should capture the channel of information, or perceptual link, that John is having with the poker-player. John’s singular thought is secured by the demonstrative-based, de re, contact with the referent, not by convention C.

John’s misidentification of the relevant player, let us call her ‘Pia’, as Claire is captured by the two different reflexive contents. If one were to embrace Kripke’s speaker’s reference/semantic reference distinction, in situations like this one could claim that the speaker, John, unbeknownst to him, “expresses” two official contents, one having the speaker’s referent as a constituent, and the other the semantic reference. John would be saying two things at once. More precisely, John’s speech act can be classified by focusing on distinct official contents, i.e. the one constrained by the name-network and the one constrained by the notion-network:

  1. (D)

    That Claire has the ace of hearts

  2. (E)

    That Pia has the ace of hearts

John’s tokening of ‘Claire’ in his perception of Pia is guided, like in the case of demonstrative reference, by his directing intention, i.e. the intention at play when one tokens a demonstrative expression. In that case ‘Claire’ works like a Russellian genuine name picking out the object the speaker, John, has in mind.

If an alert audience, realizing that John mistook Pia to be Claire, were to report what John said she could focus either in reporting (4) and say something along “John said that Claire has the ace of hearts” or on (5) and report “John said that Pia has the ace of hearts”. The first report seems to suggest (at least pragmatically) a de dicto reading, while the second a de re one, i.e. a report that could be paraphrased as: “Of Pia, John said that she has the ace of hearts”. In addressing Pia our alert reporter could say: “John said that you have the ace of hearts”. In so doing, the narrator focuses on the fact that John referred to Pia: such a report is silent on how John referred to Pia. Be it as it may, if one is faithful to Donnellan’s picture (as recently highlighted by the aforementioned authors) one is committed to the following: (i) reference depends on the having in mind and (ii) by the token of a name one fixes its reference anew. If so, then, the official content expressed by an utterance of (1) is (5), while the reflexive content is (3). On the other hand, if one subscribes to a Kripke-like picture concerning the causal chain of reference one is likely to argue that in uttering (1) John expresses proposition (4) and that the reflexive truth-conditions of (1) are represented by (2). This is the dilemma we face.

One could object that the analysis I am proposing mirrors Kripke’s speaker’s reference/semantic reference distinction, for I distinguished between talking about what one intends to talk about and referring to. If this were the case, in our scenario John talks (and thinks) about Pia while referring to Claire. We would run in what Korta & Perry characterize as a mess. Thus, what John talks about can be explained in terms of pragmatics, while what he refers to is semantic in nature. If, as Donnellan stressed, the primary function of a singular term is to single out an object of discourse one intends to predicate something about, then what one talks about and what one refers to cannot be severed. In usual situations what one talks about is what one refers to and vice versa. It is only in cases of misidentification that the two do not run in parallel because the speaker, unbeknownst to her, uses the wrong name to characterize the object she has in mind. There is a mismatch between the name-network and the notion-network. We must acknowledge this incongruity if we want to deal with our Claire-Pia scenario and to handle what Korta & Perry characterize as messes. In situations like this the having in mind, the notion-network, trumps the name-network. There may also be some sort of reference indeterminacy. Our alert audience could ask John: “Do you mean Claire or Pia?” To do so, though, our audience must grasp John’s error, viz. that he confused Pia to be Claire and, thus, that he was gaining information from Pia, the object he has in mind. To correct John’s mistake, the alert audience must first grasp the object John has in mind and intends to single out. Our alert auditor can classify what happened in using two official contents, (4) and (5). While (4) is secured by the name-network, (5) is secured by the notion-network. If, as Kaplan (2012) claims, names ride on the having in mind, in our scenario John refers to Pia. He does so, though, using the wrong label. The alert audience perceives Pia and (along with John) entertains a singular thought about her. John talks and thinks of Pia in a direct, unmediated, way. John mistakenly thinks that she bears the name ‘Claire’ (he refers to Pia by using ‘Claire’). Yet, for Pia to be the referent and the object of John’s singular thought she need not, and cannot, satisfy the property of being called ‘Claire’. This line of thought parallels Donnellan’s lesson on the referential use of definite description. With ‘the man with the Martini’ one directly refers to the relevant man, and entertains a de re thought about him (or her) even if his or her glass contains water. Thus, for the description to pick out the referent, the latter need not satisfy the descriptive content of the description, just as Pia being the referent of John’s utterance cannot satisfy the property of being named ‘Claire’.

If in the envisaged scenario the relevant player John directs his attention toward and intends to talk about were Claire (i.e., if Claire, instead of Pia, were the poker-player John was perceiving), he would have Claire in mind and we would have a unique proposition or official content, (4), to capture what he said. The having in mind and convention C would both reach the same individual and John would entertain a singular thought about Claire. Both the name-network and the notion-network would bring John’s token of ‘Claire’ to Claire.

If the picture I am presenting comes close to being the right one, we could say that the having in mind drives reference. Yet we must not reject the view that the object one has in mind often depends on the label one uses to single it out, i.e. that one has Aristotle in mind because one tokens ‘Aristotle’. We can thus argue that one’s singular thoughts are determined by the having in mind. Yet, the object one has in mind can depend both of the label one utters or the object one is directly gaining information from. Often the two links, as represented by the reflexive contents (2) and (3) converge. Yet in some awkward cases like in the Claire-Pia example they differ. We can thus argue that one’s singular thoughts are determined by the having in mind even if the having in mind relies either on the name network, as it is represented by the reflexive content (2) or the notion network, as it is represented by the reflexive content (3).

9 Conclusion

The aim of this paper was to present two distinct approaches to singular (direct) reference. Furthermore, following Donnellan’s insights, I insisted on how the notion of having in mind drives reference inasmuch as it seems to present the best way to characterize what goes on in ordinary thought episodes and communicative interactions. In focusing on cases of misidentification in the token of a proper name I suggested that Perry’s and Korta and Perry’s critical referentialism/pragmatics and the name-notion network are a good tool we can use to characterize both, how reference is fixed and the cognitive impact of utterances containing singular terms (in particular, proper names). Korta and Perry’s Critical Pragmatics, if I am right, shows us how to handle the divergences between Kripke’s and Donnellan’s viewpoints, and to deal with cases of what Korta and Perry characterize as messes.