On the Adequacy of Attitude-Ascriptions

Sobre a adequação da atribuição de atitudes proposicionais

ABSTRACT

Fregeans and Russellians defend different conceptions on the truth-conditions of reports of attitude ascriptions. Very often, they appeal to our intuitions and to pragmatic aspects to support their view. In this paper I propose a principle of adequacy of attitude-ascriptions. In the first section, I review the old puzzle concerning the failure of the principle of substitution salva veritate in attitude ascriptions. In the second section, I present my principle, which is based on the claim that by choosing a designator for an attitude-ascription a competent speaker is usually oriented by her supposition of the hearer’s epistemic state concerning the referent. In section 3, I apply my conception to some traditional tricky examples of attitude ascriptions in the literature. In section 4, I argue that even in non-standard situations (e.g. context of irony) a competent speaker must take into account the hearer’s epistemic state. Then, in the fifth section I suggest a consequence of my proposal for the discussion on exportation. Finally, in the concluding remarks I sketch some thoughts about the possibility of extending my pragmatic proposal to a semantic account of the truth-conditions of attitude-ascriptions.

Keywords:
Attitude ascriptions; exportation; contextualism; pragmatic adequacy

RESUMO

Fregianos e Russellianos defendem diferentes concepções a respeito das condições de verdade de relatos de atitudes proposicionais. Eles apelam frequentemente para nossas intuições e aspectos pragmáticos para sustentar suas posições. Neste artigo proponho um princípio de adequação de atitudes proposicionais. Na primeira seção, apresentarei o antigo enigma sobre a falha do princípio de substituição salva veritate em atribuições de atitudes proposicionais. Na segunda seção, apresentarei meu princípio, que é baseado na tese de que ao escolher um designador para um relato de atitude proposicional um falante competente é usualmente orientado pelas suas suposições sobre o estado epistêmico do ouvinte a respeito do referente. Na seção 3 eu testarei minha proposta com alguns exemplos tradicionais de atribuição de atitudes proposicionais da literatura. Na seção 4, argumentarei que mesmo em situações não-standard (p.ex. no contexto da ironia) um falante competente precisa levar em consideração o estado epistemológico do ouvinte. Então, na quinta seção derivarei consequências da minha proposta para a discussão sobre o problema da exportação. Finalmente, nas conclusões, esboçarei alguns pensamentos sobre a possibilidade de expandir minha proposta pragmática para uma concepção das condições de verdade de atribuições de atitudes proposicionais.

Palavras-chave:
Atitudes proposicionais; exportação; contextualismo; adequação pragmática

1 Substitution Salva Veritate in Intensional Contexts

Extensional contexts are usually characterized by means of some neat logical principles, including the principle of substitution salva veritate (SSV) and the principle of existential generalization (∃G), i.e.:

(SSV) interchange of co-referential expressions cannot possibly change the truth-value of a statement (the inference ‘φa, a=b, therefore φb’ is valid), and

(∃G) existential generalization (‘ja, therefore ∃x, such that jx’) is necessarily a truth-preserving inference.

Intensional contexts are contexts that lack these features-or, at least, it seems so. Here is an example of the failure of (SSV): it is certainly true that

(1) Lois Lane believes that Superman can fly.

Since Clark Kent is Superman, and given (SSV), one could expect that

(2) Lois Lane believes that Clark Kent can fly.

would also true. But since Lois Lane does not know that Superman is Clark Kent, (2) sounds false.

Fregeans¹ usually hold that an occurrence of a referential expression within the scope of a propositional attitude verb refers to the way the subject of the ascription represents the ordinary referent. Accordingly, (2) is really false, since it asserts that Lois Lane believes that the object she represents as ‘Clark Kent’ can fly. Neo-Russellians² usually claim that (2) is true despite our intuitions. For them, ‘Superman can fly’ and ‘Clark Kent can fly’ express exactly the same (Russellian) proposition, for they have exactly the same constituents organized the same way. Therefore, (1) and (2) must have the same truth-value.³

Russellian proponents of semantic accounts on the truth-value of propositional attitude reports appeal to intuitions to support their views. Some say, for instance, that (2) is true, but it ‘sounds false’ because it is pragmatically misleading. But is (2) really pragmatically misleading? More generically: is

(2) pragmatically misleading under any circumstances? Further, can we formulate a general principle of pragmatic appropriateness for propositional attitude reports? These questions constitute the main topic of this paper.

I will argue here for the claim that a pragmatic account of attitude reports must consider the epistemic state of speaker and hearer for evaluating the appropriateness of such reports. As far as I can see, current explanations ignore a general principle of linguistic empathy, according to which a competent speaker generally considers the audience’s perspective (or, at least, the perspective she attributes to the audience) in her choice of a statement’s expressions. In the particular case of attitude ascription, the speaker usually projects the audience’s perspective in her choice of the singular term used to refer to the subject of predication.⁴ Based on this principle, I shall argue that in some contexts (2) is appropriate, in others not.

Since my aim in this paper is not to provide a semantic account of attitude ascriptions I shall neither

offer a discussion on their logical form nor propose a theory of their truth-conditions. However, I think the present discussion about the adequacy of statements of attitude ascriptions may contribute to the semantics, for it provides an explanation of our ambiguous intuitions which constitute the background of the different semantic theories.

Speaker, Audience and their Epistemic State

For simplicity, I will abbreviate an assertion of attitude ascription like subject S entertains attitude Φ that α is P as:

S Φ: α is P

where α is a meta-linguistic variable for a designator (definite description or proper name) which designates object o.⁵ My diagnosis is that the traditional approach in the debate on attitude-ascription is addicted to the wrong focus on the relation between the epistemic subject S and the designator α.

In order to understand the proposal of this paper, we shall briefly recall a distinction introduced by Donnellan. In his famous paper ‘Reference and Definite Descriptions’ (1966), Donnellan introduced the distinction between the attributive and the referential use of a definite description. In his words, ‘a speaker who uses a definite description attributively in an assertion states something about whoever or whatever the so-and-so is’ (1966, p. 267) and ‘a speaker who uses a definite description referentially in an assertion (...) uses the description to enable his audience to pick out whom or what he is talking about and states something about that person or thing’ (1966, p. 267). He introduced this distinction in order to overcome the difficulties of Russell and Strawson in their debate on the evaluation of truth conditions of sentences’ definite descriptions. According to him, the main problem of both Russell (1905) and Strawson (1950) is that each focuses on just one of two different uses of definite descriptions and fails to account for the possibility of the two different uses. In a further passage, he suggests that the attributive/referential distinction is not semantic, but pragmatic: ‘In general, whether or not a definite description is used referentially or attributively is a function of the speaker’s intentions in a particular case.’ (1966, p. 272). Thus, if Donnellan is right-and I think he is-the speaker’s intentions should be considered for evaluating whether the definite description has been used in an attributive or referential sense and, as a result, for evaluating the truth-conditions of the sentence in which it occurs. Further, the speaker (at least in the attributive sense) uses the description ‘to enable his audience to pick out whom or what he is talking about’, i.e., he considers the audience’s perspective in using a designator.

In ‘Speaker’s Reference and Semantic Reference’ (1977), Kripke critically discussed Donnellan’s article (he rejected Donnellan’s claim to have refuted Russell) and argued, I think correctly, that a distinction like that between attributive and referential use of definite descriptions may be applied to proper names: If two people see Smith in the distance and mistake him for Jones, they may have the following colloquy: ‘what is Jones doing?’, ‘He is raking the leaves’. In this situation the participants of the dialogue are referring to Smith, and not to Jones. Inspired by Grice’s apparatus of the distinction between the speaker’s meaning and the word’s meaning, Kripke (1977, p. 109) suggested distinguishing the semantic referent of a designator from the speaker’s referent. When d is the usual designator for object o, o is the semantic referent of d. The speaker’s referent, in turn, is given by a specific intention, on a given occasion, to refer to a certain object, which may be other than o. In the Smith-Jones story, Jones is, of course, the semantic referent of ‘Jones’, but Smith is the speaker’s referent.

The view I wish to defend here is inspired in this distinction between semantic referent and speaker’s referent. I think we must consider the speaker’s intention in evaluating the pragmatic adequacy of an attitude-ascription and, further, this intention must be considered for both definite descriptions and proper names. My general claim is this: the adequacy of an attitude-ascription stating that S Φ: α is P do not necessarily depend on S entertaining this content via the designator used in the ascription, for usually the speaker does not intend that the subject S entertain a content about object a via the particular designator α. In the case where α is a definite description, the use of α in the ascription does not imply that S picks out the referent of α by means of this description. In the case where α is a proper name, the use of α does not imply that S picks out the referent of α by means of a special cognitive relation it holds to the referent of α (like acquaintance, causal chain, or whatever). The intuition that such implications are more or less implicitly suggested in many philosophical examples (like the Superman example) is due to the fact that such sentences are taken isolated from any context. In such artificial circumstances it sounds plausible to argue that by asserting ‘Lois Lane believes that Clark Kent can fly’ we are simply claiming that Lois Lane believes that the person she would call by the name ‘Clark Kent’ can fly. But, as I shall argue, this is not the way we normally evaluate the adequacy of attitude-ascriptions.

To develop a more plausible theory of such assertions’ adequacy, we must, first of all, distinguish three different referents: the semantic referent, the speaker’s referent and the hearer’s referent. The semantic referent of a designator is its usual linguistic referent: so John is the semantic referent of ‘John’. The speaker’s referent is the referent intended by the speaker in using a designator: as we saw, in our story, Smith is the speaker’s referent of ‘John’ (used in that particular situation). Finally, the hearer’s referent is the referent that the hearer believes that the ascriber is using the designator to refer to. In our story, both the speaker’s and the hearer’s referents coincide: both mistake Smith for Jones. So, Smith is both the speaker’s and the hearer’s referent. But this does not have to be the case. The hearer may have a different referent from the speaker in some obvious situations (in our story, if the hearer did not mistake Smith for John). Furthermore, whenever a particular assertion is addressed to more than one hearer, there may be simultaneously more than one hearer’s referent: if I say, e.g., in a conference I would love to have met my favourite philosopher, I suppose my students in the modern philosophy course will think I am talking about Leibniz, while my students in the contemporary philosophy course will think Russell is the referent of my description.

If I am right, the main criterion for deciding the adequacy of the assertion ‘S Φ: α is P’ is the speaker’s intention and his supposition of the audience’s perspective in grasping the referent of ‘α’. In a standard situation the speaker uses a designator she believes the audience will easily understand to pick out whom the speaker is talking about. Usually, the speaker’s referent, the hearer’s referent and the semantic referent simply coincide. ‘Standard situation’ is certainly a vague notion, but it may be characterized as a situation in which: (i) the speaker and the audience share common knowledge about what/who is the referent of the designator, (ii) both are aware of sharing this common knowledge, and (iii) the speaker has neither the intention of being ironic, nor the intention of deceiving the audience (e.g., by steering the audience to hold a false belief or to believe in a proposition different from the speaker’s expressed proposition), nor the intention of adding information to the hearer’s set of information about the referent. In section 4 we will see some examples of non-standard situations, i.e., situations in which (i), (ii) and (iii) fail. But, as I will argue there, even in non-standard situations our criterion for deciding the adequacy of attitude ascriptions remains valid.

Why should we consider the speaker’s supposition about the audience’s perspective decisive for deciding the adequacy of an attitude ascription assertion? I think the main argument is that this is, in fact, the way we usually reason in concrete situations. Let me illustrate this with a concrete example: suppose a couple goes to a new restaurant with their two-year-old son Augusto. As it happens, the parents were distracted for a few seconds by a conversation and did not realize that their child had gone to the toilet. After noticing Augusto’s absence, they start desperately looking for him. The parents’ worst immediate assumption is that their child has left the restaurant. Thus, the father rushes to the concierge at the door and asks him:

(3) Did you see the little boy with red hair go out?

The concierge, fortunately, denies this: ‘Certainly not, sir’. Relieved, the father tells his wife:

(4) The concierge says that Augusto did not go out.

All this sounds perfectly natural-in fact, this sounded perfectly natural for all participants in this true story. Analysis of the choice of the two different designators in (3) and (4) is fundamental for my claim.

First, concerning (3): the father’s decision to use the definite description ‘the little boy with red hair’ when asking the concierge is obviously due to the fact that, first, he did not know whether the concierge knew that ‘Augusto’ was his son’s name and, second, this description expressed a sufficiently prominent feature that any normal observer could use to identify his son. If the father were the owner of the restaurant, the members of his family would probably be known personally by name by anyone on the staff, and so the father would probably have simply asked ‘Did you see Augusto going out?’ Alternatively, if the father suspected the concierge were colour-blind he would probably avoid the predicate ‘red’ in the description. The lesson we should get: a competent speaker usually selects a designator from among several possible options for referring to an object according to the knowledge she supposes the hearer has about the referent, i.e., which terms she thinks the hearer is able to recognize as names of the object. In standard situations, the speaker selects a term following some pragmatic maxims (the most usual, the shortest, the epistemologically simpler or most salient, politically correct, etc.). Note that in most cases the speaker makes an arbitrary choice about which designator to use from among several possible more or less equally adequate designators (‘the boy with red hair’, ‘the boy wearing a yellow shirt’, etc.). Second, and more important for our purposes, when reporting the concierge’s answer to his wife with (4), the father used the proper name ‘Augusto’, because this is the usual way both refer to their son when talking with each other in everyday life. And there is no reason for changing this in this situation. By saying (4) the father is not stating or even suggesting that the concierge knew Augusto by his proper name. By using the designator α in an attitude-ascription by a subject S about a, the speaker does not always imply that S knows that α denotes a. Only this explains why the wife in the restaurant did not conclude from her husband’s report that the concierge knew Augusto by his proper name nor that her husband used this name when asking the concierge. And notice importantly, it is not always pragmatically misleading in an assertion of attitude-ascription about a subject S to use a different singular term than S used (or would use) for stating an attitude, as some neo-Russellians claim.

For this reason, I think Sosa (1970, p. 891) takes the correct path when he suggests that statements like ‘S believes x to be F’, or ‘S believes about x that it is F’, and similar statements ‘are incomplete as they stand, and are filled out by the context’. But he is wrong in claiming that the context must supply some indication of the type of singular term (name, description, indicator) “under which” S believes x to be F’. This may be true in some cases, but certainly not in all and probably not in most cases. What the context must supply is rather the type of singular term the speaker wants the audience to use for picking out the referent.

I am not claiming that the intuition behind Frege’s analysis of truth conditions of attitude ascription is wrong. Indeed, in some contexts the speaker may use the singular term she wants her audience to grasp as being that very singular term with which the believer picks out the referent. In our story, if the father thought the specific way the believer picked out Augusto contained important information for his wife, instead of saying (4) we would probably say:

(5) The concierge said that the little boy with red hair did not go out.

This would also be an adequate answer if the wife were for some reason confused by the report (4) and asked for a more accurate one: ‘what exactly did the concierge say? Did he really say that Augusto did not go out?’ The reasons for the wife being confused may vary greatly depending on the context. In any case, in such situations, in which the way the believer accesses the referent matters, the criterion of pragmatic adequacy supports Fregean truth-conditions. My only claim here is that the pragmatic adequacy not necessarily supports such truth-conditions.

It is worth stressing that the core of my proposal is not simply to consider speaker and hearer in regard to their ability to identify the referent of the attitude-ascription, but rather their mutual attribution of this epistemic state concerning this identification. In any ascription the speaker should consider the way she supposes the hearer may identify the object and, further, she should consider that the hearer will consider her own supposition when interpreting the ascription. This is important, because even Goodman (2017), who came very close to the proposal of this paper, overlooks this point in her analysis of some examples. See, e.g., her fine example

Imagine that Jim, Sally, and Daisy are members of an Arkansas bluegrass band called The Little Rock Fiddlers. Lucy knows them from her local bar, but doesn’t know that they are members of the band. But you and I know this, and you wish to report to me that Lucy believes that Jim is from Arkansas, Sally is from Arkansas, and Daisy is from Arkansas. You could use (6) to do so:

(6) Lucy believes that all of the Little Rock Fiddlers are from Arkansas.

Given that Lucy doesn’t know that Jim, Sally, and Daisy are band members (and we know that she doesn’t know), (6) cannot be interpreted as truly relating Lucy to the universal generalization that the band members are from Arkansas ( Goodman 2017 , p. 7, my emphasis).

Pace Goodman, the relevant point for understanding why this ascription can hardly be wrongly interpreted is not the fact that we both know that Lucy doesn’t know that Jim, Sally, and Daisy are band members, but rather (i) the fact that I know that you are informed about Lucy’s ignorance, and (ii) the complex reflexive fact that I know that you know that I know that you are informed about Lucy’s ignorance. Maybe easier: I know that we are mutually informed about our epistemic state concerning Lucy. If I suspected that you had no idea of Lucy’s ignorance, I would certainly not have used the description ‘all of the Little Rock Fiddlers’ to refer to them, for in this case I would be afraid you would be justified to interpret me as relating Lucy to the universal generalization. Further, if I knew that you were informed about Lucy’s ignorance, but I suspected that you did not know that I am also informed about it, I would probably avoid asserting (6), because I would be afraid you could be induced to believe I entertain a wrong universal generalized belief about Lucy. In any case, what blocks the universalized interpretation is the knowledge the speaker attributes to the hearer about the believer and about the speaker’s corresponding knowledge. In fairness, it must be said that in other passages, Goodman recognizes this point very well (though she is making a semantic claim):

In many cases, however, the way in which an object is picked out reflects the speaker’s beliefs about her addressee’s beliefs, or common knowledge’, thus ‘the truth-conditions of attitude-ascription are sometimes governed by the speaker’s beliefs about her audience’s way of thinking (or by common knowledge shared by speaker and audience) ( Goodman 2017 , p. 16).

I suppose here that by ‘common knowledge’ Goodman means my (ii): speaker’s knowledge about the audience’s epistemic state concerning the speaker’s knowledge.

We can now formulate a general principle of appropriateness for the use of singular terms in attitude ascriptions: Given the belief report ‘S Φ: α is P’ uttered in context C, constituted by speaker A and hearer H, it is appropriate for A to use α in referring to x iff (i) A knows that H is able to recognize that α is used to pick out x, (ii) the context makes it clear that by saying ‘S Φ: α is P’, A does not imply that α is a name of x for S. If condition (ii) does not hold (i.e. the context suggests that by saying ‘S Φ: α is P’, A implies that α is a name of x for S), the use of α is inappropriate.

One last word about the motivation for this paper’s claim: There is a general linguistic principle of empathy that competent speakers usually follow in communication. When a mother says to her 2-yearold son ‘Look, daddy is coming here’ she certainly does not mean that her own daddy is coming, but rather the person her son usually calls ‘daddy’. It is similar when she says, ‘Look, there is a woof-woof over there’. She usually does not refer to dogs as ‘woof-woofs’, except when talking to her son. An entomologist talking to her neighbour will probably avoid the term ‘coleoptera’, which she normally uses when discussing insects with her colleagues, and calls that bug on the wall simply a ‘beetle’. I still say ‘I got a letter from Cousin Mark’ to my 96-old father, because he has no idea what an email is. In all these cases, the speaker adjusts her word-choice according to her supposition about the hearer’s perspective. There are many ways our supposition about the hearer can influence our choice of terms: we may avoid offensive terms for a specific group, or we may use them with the conscious intention to offend them. We may avoid technical terms in order to be more accessible to the layman or, alternatively, prefer such terms when we intend to impress them. The case of the particular designator the speaker chooses in statements of attitude ascription is just one more instance of this general maxim.

3. Revisiting Old Examples

We saw that the principle (SSV) seems to fail in intensional contexts: as Fregeans argue, although Superman is Clark Kent, we apparently cannot derive

(2) Lois Lane believes that Clark Kent can fly from the uncontroversial true assertion

(1) Lois Lane believes that Superman can fly

since (2) is false. Now, is it inadequate or pragmatically misleading to utter (2)? I think it depends on the context, in particular on the knowledge of speaker and hearer about the referent and the subject of the attitude ascription. Let us distinguish three contexts C1-C3. In all of them, two colleagues of Lois Lane, A and B, are interested in her opinion about Superman’s ability, for they know that she believes that Superman is very strong, but they are not sure whether she believes that Superman can also fly.

(C1) Suppose A and B both know that they alone know Superman’s secret identity. Consequently, both know that Lois Lane does not know that Clark Kent is Superman. But most of the time A and B are accustomed to referring to Superman as ‘Clark Kent’, not as ‘Superman’. Now, one day A and B are discussing how extensive Lois Lane’s ignorance about Superman is. A believes that Lois Lane thinks Superman is very strong and that he can also fly, while B is not sure whether she thinks he is able to fly. Then, A explains to B that yesterday Lois Lane saw Superman/Clark Kent flying over the city and therefore A is sure that Lois Lane believes he can fly. In this situation, we would consider (2) not misleading because in saying it, A is not suggesting to B that for Lois Lane ‘Clark Kent’ is a name of Superman. This is just like the husband of our story, who, in saying the concierge stated that Augusto did not go out, is saying something with a true reading.

(C2) Now, suppose A knows Superman’s secret identity, but A supposes (correctly or not) that B does not know it and A has no intention to reveal it. In this case, we would not expect that, in a similar dialogue about Lois Lane’s belief in Superman’s ability, A would utter (2). The only reason for uttering

(2) on purpose would be A’s desire to reveal Superman’s secret identity to B. If A’s supposition about B’s ignorance is correct, the reaction of B to hearing A asserting (2) will probably be: ‘What?! Are you suggesting that Superman is Clark Kent?!’. On the other hand, if A’s supposition about B’s ignorance is incorrect, B’s reaction would probably be: ‘What?! Did you also know that Superman is Clark Kent? I thought I was the only one who knew this’. In any case, the hearer’s reaction reveals that she is assessing the designator’s choice in the assertion not according to what the believer (Lois Lane) believes, but according to the speaker’s epistemic state.

(C3) Suppose now that A is strongly convinced that Superman is not Clark Kent, but she knows that B believes in this identity. In fact, many times A did try hard without success to convince B that Superman was not Clark Kent, until one day she gave up trying. In order to avoid more arguments on this topic, A simply pretended she accepted B’s identity claim. In any case, both agree that Lois Lane does not think that Superman is Clark Kent and they know that both know this. Now, they are engaged in a discussion about Lois Lane’s opinion of Superman’s abilities. In this context, when A utters (2), B’s natural interpretation will be that (2) is true (insofar as B believes Lois Lane believes Superman/Clark Kent can fly). For by uttering (2), A is simply claiming that Lois Lane believes that he is able to fly. For, since what is at stake is not the question of whether ‘Clark Kent’ is a name of Superman for Lois Lane or for B, but only the issue of Lois Lane’s belief in Superman’s abilities, A may use ‘Clark Kent’ to refer to Superman. After all, she thinks B will take this name as denoting Superman, even if A herself does not think that ‘Clark Kent’ is a name of Superman. Superman is not the speaker’s referent of ‘Clark Kent’, but since she knows that Superman is the hearer’s referent for this name, she may use it to refer to Superman. At the same time, by uttering (2) A is not suggesting that ‘Clark Kent’ is a name of Superman for Lois Lane, for both know that ‘Clark Kent’ is not a name of Superman for Lois Lane. This case precludes Fregean intuition. If all this is correct, the lesson here is dramatic: It does not matter whether or not the speaker himself thinks α is a name for x, or whether or not α is a name of x for the believer (Lois Lane). What really matters is

(i) whether or not the speaker knows that the hearer recognizes α as a name for x, and (ii) whether or not by using α the speaker suggests that α is a name of a for the believer. In communication empathy is fundamental. We could imagine many other situations, but I hope this is enough for explaining our ambiguous intuitions concerning the adequacy of (2). Taken in isolation from the context, the adequacy of (2) is sub-determined. In some contexts asserting (2) may be misleading, as some Russellians claim, but not in all. And as soon as the relevant aspect of the context-the epistemic state of speaker and hearer-is fixed, the adequacy of (2) is easily decided. The main reason for supposing that an assertion of attitude-ascription like ‘S Φ: α is P’ is false in the above examples is the supposition that by saying it, the speaker is implying, or at least suggesting, that ‘α’ is a name of x for S. Whenever this supposition is not made, the assertion of attitude-ascription becomes adequate (insofar as S really Φ’s that α is P, of course).

4 Non-standard situations

In section 2, I characterized a standard situation as a situation in which: (i) the speaker and the audience share common knowledge about what/who is the designator’s referent, (ii) both are aware of sharing this common knowledge, and (iii) the speaker has neither the intention of being ironic, nor the intention of deceiving the audience, nor the intention of adding information to the hearer’s set of information about the referent. As I explained above, in such situations the speaker’s choice of the designator in an assertion of attitude-ascription is oriented by her prognosis of the hearer’s knowledge about the term’s referent. Now, I want to argue that this speaker-hearer oriented explanation is correct even in non-standard situations. This will give additional support to our view. Let us consider five non-standard situations: context of irony, context of intentional dissimulation, context of intentional compensation, context of intentional ambiguity and, finally, context of reference with addition of information.

In the context of irony the speaker uses a designator α which does not strictly apply to the intended referent o (i.e. o is not the semantic referent of α), but she supposes (correctly or not) that the audience is able to recognize that she intends to pick out object o by using α (o is the speaker’s and hearer’s referent). When the speaker’s supposition is correct and the audience picks out o, the audience grasps the speaker’s proposition. This understanding of ironic context is also valid for the evaluation of the adequacy of attitude-ascriptions. Take for example the situation of a company affair in which most of our colleagues believe that the first ex-president of our company, o1, is being unjustly accused of a crime. Only we three, you, our friend Peter and I think that o1 is guilty and will be justly arrested. More importantly, we three know of each other that only we three think this way and we very often secretly tease our colleagues because of their naïve opinion about o1. They think (wrongly, we would say) the second ex-president o2 is guilty. This morning Peter tells me he believes that ‘α will be arrested tomorrow’ (where α is any designator of o1). I may report this to you:

(7) Peter believes that the poor innocent ex-president will be arrested tomorrow.

Since you know that none of us three believes o1 is really innocent, you will recognize the irony and pick o1 as the referent of ‘the poor innocent ex-president’. More important for our point: you know that Peter himself does not consider this description correct for referring to o1 (in the strict sense of ‘correct’, i.e., besides context of irony) and therefore you will properly understand that either Peter used this description in the sense of irony or he used another singular term, which I replaced with the description in my report only for reasons of irony. In any case, in my decision about which designator I should use in (7), I considered your epistemic state about S, me and o1. If I thought you were unable to decipher the intended referent of ‘the poor innocent ex-president’ or if I thought you could be led to a misconception of Peter’s real opinion about o1, I would have avoided irony.

In the context of intentional dissimulation, the speaker’s referent and the hearer’s referent of a designator are different. Here the speaker uses a singular term α, such that a is its referent for the speaker and b (with a≠b) its referent for the hearer-and the speaker knows this. Thus, the speaker uses α on purpose in order to lead his audience to believe that she is talking about b and not a. Let us consider an example. Suppose again the same company affair context. But now, one of our naïve colleagues who believes in the innocence of o1, viz. John, asks me whether I know some news about the company affair. Since I wish to avoid quarrels with John and do not want him to discover Peter’s and my true opinion, I may say, again:

(7) Peter believes that the poor innocent ex-president will be arrested tomorrow.

Here my intention is obviously not simply irony, but dissimulation. In fact, I was ironic myself (I called the man ‘innocent’ who I believe not to be innocent), but I knew the hearer would not realize that I was being ironic. If I am successful in my intention, John will grasp the Russellian proposition that o1 will be arrested tomorrow. Further, John will be led to the false belief that Peter and I consider o1 innocent. For this reason, some will probably think that in this context this assertion is a lye. Anyway, as in the previous scenario, the speaker selects the designator for the ascription report according to her belief about the audience’s epistemic state.

Another non-standard context is that of intentional compensation. If your hearer is your elderly grandfather, and you know he has no idea what an email is, you may say

(8) Father told me Uncle Ben’s last letter was quite sad

although strictly speaking what your father told you is that Uncle Ben’s last email was sad. You just select the designator ‘Uncle Ben’s last letter’ in this context, because you think that if you say this the audience will more easily understand your utterance. Again, the speaker selects the designator according to her supposition about the hearer’s perspective.

In the context of intentional ambiguity the speaker uses a singular term α when talking to two (or two groups of) hearers H1 and H2, because she believes that, by using α, H1 will pick out object o1, while hearer H2 will pick out a different object o2. Different hearers’ referents are at stake at the same time. Here is an example: Suppose a father got only two tickets to a soccer game and is planning to take one of his two sons with him. The two sons are discussing who is going to accompany their father and ask their mother whether she knows their father’s decision. Wise as she is, in order to avoid further discussion for the moment, she simply says

(9) Father told me he is going to take the most well-behaved son.

Since each son considers himself the most well-behaved son, each thinks he will be chosen. This way, the two hearers will grasp two different Russellian propositions. Note that in this case the speaker knew that different hearers would have different referents associated with the designator ‘the most well-behaved son’ and made use of this duplicity on purpose. Again, the audience understood the speaker’s choice of the designator. Of course, we may create similar but more complex situations in which one assertion may express three or more propositions according to the number of hearers and their different beliefs.

We may imagine many different situations, but these four examples should be enough to make my point clear, viz. that even in non-standard situations, the speaker’s choice of the singular term in an attitude-attribution depends directly on her knowledge of her audience’s epistemic state and the proposition she wants it to grasp with this assertion. Further, a competent hearer will usually understand the speaker’s intention by interpreting her ascription. In many cases, the hearer will not associate the designator used by the speaker to attribute a propositional attitude to S with B’s epistemic state, but rather with the speaker’s epistemic state or with the common epistemic state of speaker and hearer.

There is, however, at least one non-standard situation in which the speaker does not seem to select the designator for the attitude ascription according to her supposition about the hearer’s beliefs. This is the context of reference with addition of information, i.e. the context in which the speaker has the intention to refer to and inform the hearer about the referent of the designator. I will borrow here an example from Kaplan (1968-69). (Let me stress that I am making here a quite different issue as that which motivated Kaplan to this example.) Imagine that Jack has recommended a mechanic he considers to be honest. Now, you have discovered that the mechanic is dishonest: he uses stolen parts and charges for work not done. You can accurately report to me in these terms:

(10) Jack thinks that the lying SOB who fixed my car is honest.

What you mean by asserting this is certainly not that Jack thinks a lying SOB is honest, but rather that Jack considers the mechanic, whom the speaker believes to be a lying SOB, to be honest. In this example, the speaker does not select one designator she considers the hearer can most easily use to pick out the referent. In fact, the hearer supposedly never heard about that mechanic. The speaker introduced the definite description ‘the lying SOB who fixed my car’ exactly on the grounds that she wanted to refer to and inform the hearer about the referent. This could appear as a counter-example to my claim that, in general, the speaker selects the designator according to her supposition about the hearer’s beliefs, since in this case the hearer did not believe anything about the referent.

However, I think, correctly understood, this is really not a counter-example. For even in this case, what guides the speaker in her choice of the designator is her knowledge of the hearer’s epistemic state: since the speaker believes the hearer did not know about the mechanic’s dishonesty, and since she considers this an important piece of information for the hearer in grasping the intended content of the assertion, she introduced the referent with this definite description, adding in this way some relevant information. The difference between this and the previous contexts is that in those contexts the speaker selected the singular term according to her supposition about how the audience effectively thinks about the referent, while in this latter context what matters is how the speaker thinks the audience should think about the referent.

5 Puzzle of Exportation

The principle of empathy with the audience that I have been proposing here may help explain some tricky issues on exportation. Obviously, the topic of exportation and singular thought is too complex to be discussed adequately in a single section. In this section, my purpose is to suggest some devices for re-evaluating the debate on exportation.

Quine (1956) influenced the contemporary debate by introducing a quite elegant way to distinguish two readings of the ambiguous sentence:

(11) Ralph believes that someone is a spy

which may mean that

(12) Ralph believes of someone, that he is a spy.

or, alternatively, that

(13) Ralph believes that there are spies.

In ordinary terminology, we say that while (12) is a de re statement of the form ‘∃x (Ralph believes x is a spy)’, (13) is a de dictum statement of the form ‘Ralph believes ∃x (x is a spy)’. Logically, (12) implies

(13) but not vice-versa. The fact that we naturally understand this distinction suggests that we do entertain so-called ‘singular thoughts’, i.e., thoughts about an object that is not picked out by means of a qualitative description, but ‘directly’.

Exportation is defined as the inference, via existential-generalization, from a de dicto formulation over a singular term within the ‘that’-clause, to a relational formulation de re, whereby the singular term is put outside the context ‘believes that’, allowing free substitution of co-designative singular terms.

However, many problems emerge at this point. The first, pointed out by Quine (1956), is that quantification into intensional contexts leads to the result that we may impute irrationality to rational and competent speakers: just suppose Ralph knows Orcutt and does not believe he is a spy. But then Ralph sees a suspect concealed man and thinks immediately ‘that man is a spy!’, without noticing that that man is Ortcutt. Thus, we could attribute to Ralph the contradictory de re belief about Ortcutt that he both is and is not a spy. From this (and other reasons we can ignore here) Quine derives his scepticism about the intelligibility of de re thoughts.

Kaplan (1968-69) and Sosa (1970) also contributed to this debate. In ‘Quantifying In’ Kaplan (1968-69) proposed a criterion for deciding when an exportation is sound. To do this, he offered a more perspicuous definition of de re thoughts. According to him, de re thoughts may be characterized by being expressed with a term α that represents an individual x in a special way for the believer B. More exactly: α represents x to the believer B (‘R(α, x, B)’) if and only if (i) α denotes x, (ii) α is a name of x for B, and (iii) α is sufficiently vivid (Kaplan, 1968-69, p. 203). In this way, when believer B possesses the name α, which is a vivid name of an object x, B may entertain a singular thought about x.

According to Sosa (1970, p. 888), however, Kaplan’s proposal is too restrictive, for it does not permit singular thoughts about future objects, since he requires the genetic character of the name, i.e. a causal chain of events leading from the name to its production. Further, Sosa (1970, p. 890 ff.) defended unrestricted exportation pointing to some cases in which we consider the inference of a de re statement from a de dictum statement to be sound: Suppose a sergeant, after consulting his superiors, tells the shortest man in his platoon ‘Shorty, they (the higher authority) want you to go first’. Since the superiors do not know Shorty personally, what they strictly said was only that they want the shortest man in the platoon to go first. Therefore, in this situation we go from the de dictum ascription

(14) The higher authority wants the shortest man to go first

to the de re statement

(15) Shorty, they (the higher authority) want you to go first.

Why is exportation fully acceptable in this context? Sosa (1970, p. 890) discussed many such examples and concluded that the reasons for justifying exportation are a wholly pragmatic matter, sometimes it is very definite, sometimes indefinite: the ‘context-sensitivity of exportation is no greater than that of demonstrative reference or ambiguities of predication’ (1970, p. 895). These aspects can change radically according to the context and cannot be rigidly regimented.

In fact, I think we can use our principle of empathy with the audience to illuminate the debate. According to my proposal, despite appearances and Sosa’s opinion, there is no exportation at all in this example. For (15) is not, as its superficial form may suggest, a de re ascription. Just as in the father’s report to the mother, ‘the concierge said that Augusto did not go out’, in (15) the hearer is not entitled to conclude from the speaker’s assertion that the believer (the concierge or the higher authority) entertains a de re belief. In both cases, the hearer will understand that the use of the proper name in the ascription is not the speaker’s device for ascribing a singular thought to the believer. The hearer’s background information is similar in both cases: the mother knows that the concierge does not know Augusto by his proper name, just as Shorty knows that the higher authority does not know him by the name ‘Shorty’. For this reason, in both cases the hearer will not conclude that the speaker had the intention of attributing a de re belief to the subject of ascription.

Similar reasoning also applies to the example proposed by Stalnaker (2009): Kaplan is going to a conference where he is the unannounced speaker. Since his taxi is late, Kaplan’s wife calls the event’s organizers to inform them that he will be a bit late. After calling, she reports to him: ‘you can relax, the audience knows you will be late’. The occurrence of ‘you’ in a statement usually insinuates a de re reading, for the speaker is talking directly to the pronoun’s reference. But in this case, Kaplan will certainly not interpret this statement as a de re statement about him. Of course, the wife could have reported the same saying ‘the audience knows that the unannounced speaker will be late’, but this is not necessary, given the context: the speaker (wife) knows that the hearer (Kaplan) knows that S (the audience of the conference) does not entertain a de re belief about Kaplan.⁶ Once more, the speaker’s choice of the designator is oriented by the hearer’s perspective.

If I am right, our intuitions about the adequacy of an attitude report change immediately as soon as we change the context. Suppose that Shorty is a soldier and, at the same time, a very famous opera singer. Further, everyone, including the sergeant, the higher authority and Shorty himself knows him by his stage name ‘Shorty’ and, equally important, they all know this is public information. In this case, when hearing (15), Shorty will be justified in thinking that in fact the higher authority has a de re belief about him. Or suppose that Kaplan and his wife know that only the event’s organizers still believe the identity of the unannounced speaker is a secret, for they informed the audience in advance that he would be the unannounced speaker. In this case, Kaplan would be justified in believing that his wife’s statement should be interpreted as a de re belief about him.

Conclusion: Prospects for a Semantic Account

I argued so far that traditional explanations of attitude reports are mistaken, insofar as they do not consider an essential contextual aspect for evaluating the adequacy of these statements. According to my diagnosis, current explanations ignore a general principle of linguistic empathy, according to which a competent speaker usually chooses the expressions of a statement according to the audience’s perspective (or at least according to the perspective the speaker attributes, correctly or not, to the audience). In the particular case of attitude ascription, the speaker usually projects the perspective of the audience when choosing the designator. Therefore, in evaluating the adequacy of an attitude ascription like S Φ: α is P we should not focus on the relation between S and the designator α, but rather on the epistemological state of the speaker concerning the audience’s perspective. It is neither necessarily nor usually true that by asserting that S holds attitude Φ about α being P, the speaker is suggesting that S picks out the referent of α by means of α.

Can my proposal on the adequacy of attitude reports grounds a semantic theory, in particular, an account of their truth-conditions? I think that the answer is positive or, at least, that there are good prospects of such a semantic theory. After all, like Soames (2008, p. 460), I think a semantic theory should provide ‘a rational reconstruction of the ability of speaker-hearers to identify what is asserted and conveyed by an utterance’-and this is exactly what is provided in my pragmatic account. And like Donnellan I think the speaker’s intention should be considered for evaluating the truth-conditions of a statement. In fact, the passage quoted from Goodman (2017, p. 16) above suggests an account of the truth-conditions in this line, to recall: ‘the truth-conditions of attitude-ascription are sometimes governed by the speaker’s beliefs about her audience’s way of thinking (or by common knowledge shared by speaker and audience).’

Of course, this possible extension to semantics has to face a challenge. Kripke (1977) recognizes that the distinction between the attributive and referential use of definite descriptions made by Donnellan is real, and that it is of fundamental importance for understanding some pragmatic aspects of language. But he rejects the claim that this distinction grounds a satisfactory explanation of truth conditions of sentences. Kripke proposes a situation in which someone sees a woman with a man. Taking the man to be her husband, and observing his attitude towards her, he says ‘Her husband is kind to her’ and someone else agrees, ‘Yes, he seems to be’. But suppose that the man in question is not her husband, but her lover, to whom she has been driven precisely by her husband’s cruelty. According to Russellian and Fregean analysis, these sentences are false. But according to Donnellan, the participants of the dialogue are saying something true, since they are referring to the kind lover. However, and this is Kripke’s criticism, “what if we ask: ‘Yes, but was the statement he made true?’ Donnellan would hedge. For we are not under the misimpression that the man the speaker referred to was her husband, we would not express the same assertion by ‘her husband is kind to her’’’ (Kripke 1977, p. 107, his emphasis). Thus, Donnellan may be right about our intuition concerning reference-but is this enough for grounding an adequate theory of truth conditions? Kripke seems to be appealing here to our intuition that, for grounding a theory of truth conditions we should not observe the speaker’s or hearer’s referent, but rather the ‘objective’ semantic referent. Someone could use the same line of reasoning against our proposal: Is the statement ‘Lois Lane believes that Clark Kent can fly’ really true? Or: ‘did the concierge really say that Augusto did not go out?’

In fact, I think Kripke’s criticism of Donnellan does not affect the proposal of expanding our results to a semantic theory. For our suggestion is not simply to ground the truth-conditions of statements on the basis of the distinction between referential and attributive use of designators, but rather to consider the speaker’s perspective and her beliefs about the hearer’s beliefs in specifying the truth-conditions of assertions of attitude-ascriptions. Kripke invites us (readers of his paper) to consider from our perspective: do we think the statement ‘her husband is kind to her’ is true? Indeed, we would hedge. But I would hedge exactly because I (the author of this paper) wouldn’t say to you (my and Kripke’s readers) ‘her husband is kind to her’ since I know that ‘that man’ is not her husband and because I know that you know the same thing I know. As soon as the context (speaker and hearer) of an assertion changes, its truth conditions change.

Thus, if we both (you, the reader, and I) know that Lois Lane does not know that Superman is Clark Kent, and I know that we both know that we know this, I can truly and confidently say to you that Lois Lane believes Clark Kent can fly. After all, truth conditions must be determined once the referents of the statement’s constituents are determined.

References

• BEALER, G. 1993. Solution to Frege’s Puzzle. Philosophical Perspectives, 7: 17-60.
• BRAUN, D. 1998. Understanding Believe Reports. The Philosophical Review, 107(4): 555-595.
• BRAUN, D. 2002. Cognitive Significance, Attitude Ascriptions, and Ways of Believing Propositions. Philosophical Studies, 108: 65-81.
• DONNELLAN, K. 1966. Reference and Definite Descriptions. Philosophical Review, 77: 281-304.
• DORR, C. 2012. Transparency and the Context-Sensitivity of Attitude Reports, in M. GARCIA-CARPINTERO; G. MARTÍ (Eds.), Thinking and Talking about Nothing, Oxford, Oxford University Press.
• FORBES, G. 1989. Languages of Possibility Cambridge, Basil Blackwell.
• FORBES, G. 1990. The indispensability of Sinn. Philosophical Review, 99: 535-563.
• FORBES, G. 1993. Reply to Marks. Philosophical Studies, 69: 281-295.
• GOODMAN, R. 2017. Do Acquaintance Theorists Have an Attitude Problem? Australasian Journal of Philosophy, 96(1): 67-86.
• GRICE, P. 1961. The Causal Theory of Perception. Proceedings of the Aristotelian Society, 35: 121-153.
• KAPLAN, D. 1968-69. Quantifying In. Synthese, 19(1-2): 178-214.
• KRIPKE, S. 1977. Speaker’s Reference and Semantic Reference. Midwest Studies in Philosophy, II, p. 255-276. Reprint in French, UEHLING; WETTSTEIN (Eds.), Contemporary Perspectives in the Philosophy of Language, Minneapolis, University of Minnesota Press, p. 6-27.
• KRIPKE, S. 1988. Puzzle About Belief. In N. SALMON; S. SOAMES (Eds), Propositions and Attitudes New York, Oxford University Press.
• PIETROSKI, P. 1996. Fregean innocence. Mind and Language, 11: 338-70.
• QUINE, W. V. O. 1956. Quantifiers and Propositional Attitudes. Journal of Philosophy 53:177-87.
• RUSSELL, B. 1905. On Denoting. Mind, 14: 479-493.
• SALMON, N. 1986. Frege’s Puzzle Cambridge, MA, MIT Press.
• SALMON, N. 1992. Reflections of Reflexivity. Linguistics and Philosophy, 15(1): 53-63.
• SAUL, J. 2007. Simple Sentences, Substitution, and Intuitions Oxford, Oxford University Press.
• SCHIFFER, S. 2003. The Things We Mean Oxford, Oxford University Press.
• SCHIFFER, S. 2016. Replies to the Essays. In G. OSTERTAG (Ed.), Meanings and Other Things: Themes from the Work of Stephen Schiffer, Oxford, Oxford University Press.
• SOAMES, S. 2008. Drawing the Line Between Meaning and Implicature - and Relating both to Assertion. Noûs, 42: 529-554.
• SOSA, E. 1970. Propositional Attitudes de Dictu and De Re. The Journal of Philosophy, 67 (21): 883-896.
• STALNAKER, R. 2009. What Is De Re Believe? In J. ALMOG; P. LEONARDI (Eds.), The Philosophy of David Kaplan Oxford, Oxford University Press, p. 233-45.
• STRAWSON, P. F. 1950. On Referring. Mind, 259 (235): 320-344.

• ¹
  See for instance Forbes (1989, 1990, 1993) and Pietroski (1996).
• ²
  For a defense of naive Russellianism, see David Braun (1998, 2002), George Bealer (1993), Nathan Salmon (1992)
• ³
  For different versions of Neo-Russellianism see Salmon (1986), Braun (2002) and Saul (2007). For a recent analysis see Dorr 2012.
• ⁴
  Although the arguments of this paper are, as far I can see, original, in some respects the main claim echoes Goodman (2017) and Schiffer (2003, 2016).
• ⁵
  Although this abbreviation resembles “S Φ that: α is P”, which is very often used by the philosophers of language (e.g. Kripke, 1988) to indicate the de dicto interpretation of the ascription, I do use it in a neutral sense (neither the de dicto nor the de re interpretation is intended here). Thanks for an anonymous referee for this remark.
• ⁶
  Similar reasoning also applies to the Sosa’s spy example (1970, p. 891) and many others in the literature.

Publication Dates

• Publication in this collection
  02 Sept 2022
• Date of issue
  2022

History

• Received
  29 Dec 2021
• Accepted
  13 Apr 2022