1. Introducing the Duck
In his Philosophical Investigations, Ludwig Wittgenstein wrote that “the importance of the true confession does not reside in its being a correct and certain report of a process. It resides rather in the special consequences which can be drawn from a confession whose truth is guaranteed by the special criteria of truthfulness” (1953, p. 222). In two earlier papers I argued that when the contrast between “reporting” and “confessing” thoughts is properly understood, Wittgenstein was right to hold that sincerity (“truthfulness”) guarantees the truth of the confession.Footnote 1 I assumed without argument that the resulting “expressivist” explanation of this guarantee is also an explanation of the philosophically puzzling phenomenon of first-person authority. A. Minh Nguyen takes me to task for this further assumption, and so compels me to address matters that were evaded in those earlier papers.Footnote 2 On his view my argument might successfully forge a connection between the truth of (at least some) self-ascriptions of mental states and their sincerity but, even so, what it explains is not first-person authority.
Nguyen concludes by quoting Crispin Wright’s assessment of the expressivist account of first-person authority: “The expressivist proposal flies rather further than is usually thought. But it is a dead duck all the same.”Footnote 3 Perhaps philosophers can be forgiven for not knowing that dead ducks fly no distance at all. In what follows I briefly rehearse the expressivist explanation, and then attempt to deflect Nguyen’s shot at the duck.
2. The Anatomy of the Duck
The expressivist account of first-person authority takes the form of an argument which, if successful, both explains the phenomenon and vindicates the thought that we do have first-person authority. Only three premises are needed:
(1) A speaker’s utterance of a self-ascription is true if and only if the speaker has the mental state her utterance ascribes.Footnote 4
(2) A speaker’s utterance of a self-ascription is sincere if and only if the speaker has the mental state her utterance expresses.
(3) Typically, a speaker’s self-ascription expresses the first-order mental state ascribed by her utterance.Footnote 5
From (2) and (3) it follows that
(4) Typically, a speaker’s self-ascriptive utterance will be sincere if and only if she has the first-order mental state ascribed by her utterance.
But from (1) and (4) it follows that
(5) Typically, a speaker’s self-ascriptive utterance will be true if and only if it is sincere.
The truth of an utterance that expresses the state it ascribes is guaranteed by its sincerity, and so we have Wittgenstein’s “confession whose truth is guaranteed by the special criteria of truthfulness”.
The distinguishing feature of expressivist accounts of first-person authority appears in premise (3), which states the thesis of avowal expressivism. This tells us that the sentence “I want a drink” typically is used to express a desire for a drink, and “I believe that p” typically expresses a belief that p. Avowal expressivism does not deny that these same sentences can be (and often are) uttered to express second-order beliefs about the first-order mental states they attribute. But with the qualification “typically,” the account is betting that our use of self-ascriptive utterances to express our first-order mental states is as common as our authoritative self-ascriptions of those states. A familiar but contrasting view (henceforth, avowal cognitivism) says that self-ascriptions are always employed to express second-order beliefs about first-order mental states. Adopting Nguyen’s idiom, I will refer to those utterances said by (3) to be typical as “expressive self-ascriptions,” reserving “reports” for self-ascriptive utterances that express second-order beliefs.
Avowal expressivism is widely regarded as an apt pre-philosophical description of a range of utterances with the surface syntax and semantics of first-person psychological descriptions. But it has been thought irrelevant to the explanation of self-knowledge or first-person authority since, it was presumed, utterances that are employed to express feelings, desires, hopes, and wishes cannot be truths. To have first-person authority is, at a minimum, to have the ability to reliably and effortlessly say true things about oneself. But only those self-ascriptions used to express second-order beliefs about mental states are capable of bearing truth-values or expressing knowledge.Footnote 6
Two doctrines – the autonomy of meaning, and minimalism about truth-evaluability – jointly suggest that this was a mistake.Footnote 7 The autonomy of meaning from illocutionary force allows that a given sentence might be uttered in the performance of a variety of different speech acts without change of syntax or meaning.Footnote 8 But different acts of speech can express different mental states: requests express desires, promises express intentions, and assertions express beliefs. A minimalist criterion for truth-evaluability then says that the aptness of a meaningful sentence for truth or falsity is settled by features of its surface syntax and inferential role.Footnote 9 But self-ascriptions have the needed features and, so, an utterance of “I want a drink” will be true or false whether the sentence is employed to express a belief about one’s desire, or to make a request, and so to express a desire. The combination of the autonomy of meaning with the minimalist criterion for truth-evaluability results in the idea that the mental states expressed by utterances of a sentence can vary even while the syntactic and semantic features of the sentence remain invariant. Expressive self-ascriptions are thus reconciled with truth-aptness, and accounting for first-person authority becomes trivial.
The account does not make us infallible self-ascribers, but it delimits the scope for error. First, our self-ascriptions do not always express the states they attribute – they sometimes express only second-order beliefs about the states they ascribe. Secondly, a speaker may not have the state she expresses, and so her utterance may not be sincere. Thirdly, it is not always a determinate matter that we have or lack a given mental state – further deliberation or resolve may be needed to fix the content of a still-vague thought.Footnote 10 Finally, (1) – (5) still permit us to contest the truth of a self-ascription by contesting the speaker’s grasp of meaning.Footnote 11
Discounting such errors, there is now a class of self-ascriptions – those expressing the very states they ascribe – for which sincerity guarantees truth. We thus have an important asymmetry between first-person present-tense ascriptions and other-person other-tense ascriptions (for which sincerity never provides a guarantee of truth). Nguyen’s objection to this account of first-person authority – to which I turn next – does not appear to doubt any of this. What he doubts is that in explaining this much, what I have explained is first-person authority.
3. The Duck Hunter
Nguyen’s argument turns on a conception of first-person authority he calls the “standard characterization” (SC). According to that characterization,
(SC): [I]t is necessary that, for any person, if he sincerely ascribes the presence, or absence, of a particular intentional state to his present self, then there is a legitimate presumption that what he says is true. (104)
As Nguyen notes, (SC) makes the existence of first-person authority “a deeply noncontingent fact” (126). He sketches one widely received form of support for this: namely, that the standard characterization captures the thought that first-person authority “is constitutive or definitive of mental states, or of minds, or of concepts of these” (Shoemaker), or – somewhat more modestly – that “Wholesale suspicion about my attitudinal avowals – where it is not a doubt about sincerity or understanding – jars with conceiving of me as an intentional subject at all” (Wright).Footnote 12 If it is true that first-person authority is a mark of the mental, then necessarily, if any subjects have mental states, they have states of just the sort that are suited for authoritative self-ascription. It does not yet follow that they will, or even can, authoritatively self-ascribe the states that are thus suited. The unreflective beasts of the field have mental states, but it seems very unlikely they can be accorded first-person authority. There may also be conceptual deficits or cognitive impairments that render a person incapable of self-ascriptive thoughts or utterances. But inspired by this view of the mental, we should suppose that necessarily, if subjects both have and self-ascribe mental states, then they will have first-person authority.
According to Nguyen, the problem with the expressivist account sketched in the preceding section is that it cannot provide for the necessity in (SC), and so leaves it a merely contingent matter that self-ascribers have first-person authority:
The real objection is that the typicality of expressive self-ascriptions that Jacobsen alleges is at best contingent. Such contingency may explain why self-ascriptions are authoritative as a matter of fact, but it cannot explain why authoritativeness is a necessary feature of self-ascriptions. In default of this further explanation, first-person authority remains unexplained. (Nguyen, 115)
The argument for this is straightforward: There is a logically possible world – called D-world – in which expressive self-ascriptions are either atypical or nonexistent. So on D-world, self-ascriptive utterances either typically or always report first-order mental states by expressing second-order beliefs. As Nguyen tells us,
D-worlders never express their propositional attitudes by saying that they have them. They do not get to possess, let alone exercise, this mode of competence. … [T]hey are nonetheless capable of expressing these states by standard alternative means. … For instance, a D-worlder can use “p” to express his belief that p, and use “I believe that p” to express his belief that he believes that p, though he cannot use the latter to express his belief that p.Footnote 13 (118)
But since D-worlders – like all subjects who both have and self-ascribe mental states – have first-person authority, and since D-world self-ascriptions are rarely (or never) expressive, the considerations in (1) – (5) cannot explain the authority they have. Alternatively, if expressivism provides the right account of first-person authority, then D-worlders must lack first-person authority, and so it cannot be a necessary feature of self-ascriptions. So either expressivism fails to explain first-person authority, or – contrary to (SC) – it makes first-person authority contingent.
4. The Duck Lives
Nguyen takes great care to argue that D-world is a possible world, and I have no quarrel with this. But I think there are reasons to be wary of the formulation of (SC), about which I have two reservations. First, what (SC) tell us is necessary is that there be a legitimate presumption that what a person says is true whenever she sincerely self-ascribes a mental-state. But it cannot be necessary that there be any such presumption. For there is also a possible world (F-world) in which the Freudian memes from early twentieth-century Vienna have triumphed. In F-world, everyone believes that self-ascriptions always involve wishful thinking, delusion, and self-deception. F-worlders believe that only those few who have undergone years of arduous analysis can even begin to self-ascribe their real thoughts and feelings. So no one in F-world would simply presume that a self-ascription is true, even knowing that the speaker is sincere. If F-world is logically possible, then it is not necessary that there be any such presumption. But perhaps the point of (SC) was only that it is necessary that if anyone were to presume that a sincere self-ascription is true, then her/his presumption would be legitimate. But even that is too strong. What is legitimate for any person to presume is a function of the other beliefs that person already holds. But given the radical Freudian background beliefs of F-worlders, in light of which any presumption could be legitimate for them to make, no such presumption would be legitimate for an F-worlder to make. So it cannot be necessary that there be any such presumption.
The preceding difficulty arises only because (SC) places the presumption that a sincere self-ascription is true within the scope of the necessity operator, and so makes the existence of the presumption a necessary constituent of first-person authority. I have no reservations about the claim that we legitimately presume that sincere self-ascribers are right about themselves. But if we stopped being presumptuous, subjects would not lose any of the authority they have about their own mental states. The task of an explanation of first-person authority is not to explain why anyone presumes that sincere self-acriptions are guaranteed to be true. The task is to explain how there could be any such guarantee.
For this reason, I suggest a tentative characterization of first-person authority that removes the notion of presumption from the scope of the necessity operator. But my characterization will go further still, in order to focus the central disagreement between Nguyen and myself, which concerns the status of avowal expressivism. There are two steps needed to extract a characterization of first-person authority directly from (1) – (5) of the expressivist account. First, whether or not avowal expressivism is true, if (1) – (5) is a valid argument, then necessarily, if (1) – (3), then (5). This gives us the desired “necessity,” which is now straightforward logical necessity. Secondly, the “typically” running from (3) to (5) in the expressivist account serves only to limit any connection between truth and sincerity that is earned by the argument to those self-ascriptions that express the mental states they ascribe. We can make this restriction explicit by re-writing the claim that necessarily, if (1) – (3), then (5), as: expressivist characterization (EC): If (and only if) a person’s self-ascription expresses the mental state it ascribes, then, necessarily, her self-ascription is true if and only if it is sincere.
On this unabashedly expressivist characterization, it will be legitimate to presume that a speaker’s sincere self-ascription is guaranteed to be true only if it is also legitimate to presume that her self-ascription expresses the state it ascribes. Whether there is any legitimate presumption to this effect is a contingent matter, but it is not contingent that persons who do make sincere expressive self-ascriptions will make true self-ascriptions. This way of characterizing first-person authority will surely be seen as question-begging, so I will offer a defense of it before concluding.
My second reservation about (SC) – and this applies to (EC) as well – is that it fails to capture one common aspect of the use to which the notion of first-person authority is put, even by Nguyen. Nguyen sometimes predicates first-person authority of self-ascriptions, as when he writes “[A]ll self-ascriptions carry authority in the sense of enjoying a rebuttable presumption of truth” (124). But he also predicates first-person authority of subjects, as when he writes about “the authority that every mature human subject enjoys with respect to his own present intentional states” (122). Nguyen is not equivocating – he is making use of a perfectly good distinction, but one we need to mark explicitly. We should distinguish the claim that self-ascriptions are authoritative, where that means that their truth is guaranteed by their sincerity, and the claim that subjects have first-person authority, where that means that subjects have the capacity or ability to make authoritative self-ascriptions. Clearly, if a speaker makes authoritative self-ascriptions, then she has the capacity to do so, and so she – qua subject – has first-person authority. But the converse is not true. A speaker with the capacity to make authoritative self-ascriptions might not actually do so, and so it is possible that at least some subjects who enjoy first-person authority might not actually make any authoritative self-ascriptions. In light of the distinction between authoritative self-ascriptions and authoritative subjects, it will not follow that persons in D-world do not enjoy first-person authority even if they never exercise that authority by uttering authoritative self-ascriptions.
The importance of the preceding point becomes clear when we recall that our only reason for insisting that D-worlders have first-person authority was that first-person authority is a mark of the mental, where that implies that any subject both having and self-ascribing mental states must have first-person authority. But the idea that authority is a mark of the mental does not imply that such subjects will actually make any authoritative self-ascriptions. Nor, contrary to (SC), does it imply that every sincere self-ascription will enjoy a legitimate presumption of truth. Nguyen tells us that “the expressivist proposal is doomed if expressive self-ascriptions are anything but typical” (118), and “the expressivist scheme is unworkable if expressive self-ascriptions are anything but normal” (121). But even if expressive self-ascriptions were extremely rare on D-world, D-worlders will, a fortiori, have the capacity for authoritative self-ascriptions and, so, they would have first-person authority. So the logical possibility of D-world does not put expressivism into conflict with the idea that first-person authority is a “deeply noncontingent fact” about subjects.
5. The Duck Quacks Back
Nguyen’s objection to the expressivist account was that it could not accommodate first-person authority for logically possible D-worlders, and so it left first-person authority merely contingent. I have argued that expressivism is fully consistent with the idea that first-person authority is a noncontingent fact about subjects. But it is one thing to show that the hunter’s shot has missed the mallard, and quite another to show that the duck is healthy.
In responding to Nguyen’s objection, I acquiesced to the consequence that D-worlders rarely (or never) actually make authoritative self-ascriptions, and my only ground for this was that D-worlders generally abstain from using self-ascriptions expressively. Along the way, I recommended a characterization of first-person authority according to which there is a guarantee – a necessity – that a sincere self-ascription be true, but only if it is an expressive self-ascription. Short of begging-the-question, is there any reason for insisting on this point? I think there is.
A necessary condition for the truth of a self-ascriptive utterance to come apart from its sincerity is that the state it attributes be distinct from the state it expresses. For the avowal cognitivist, this condition is always satisfied. Self-ascriptions will then answer for their truth to one mental state, and answer for their sincerity to another mental state. Assuming only that these distinct states are logically independent – that is, that it is logically possible to have the one without the other – then there will always be a possible world in which truth and sincerity do not coincide. So if avowal cognitivism were true, there could be no necessity that if self-ascriptions are sincere, then they will be true. Avowal cognitivism ensures that any coincidence of truth and sincerity in self-ascriptions will be contingent. But the claim of avowal expressivism is that, for an important subset of self-ascriptions, the state expressed is the state attributed. Such self-ascriptions answer for their truth to the very state to which they answer for their sincerity. When that is so, truth and sincerity simply cannot come apart. So only expressivism can provide the necessity that Nguyen requires for authoritative self-ascription.
In his closing reflections, Nguyen writes that “Jacobsen has never given, or attempted to give, an argument for [avowal] expressivism. …” and “In the absence of a cogent argument for [avowal] expressivism … a sceptic is surely justified in asking how we know that [avowal] expressivism obtains” (p. 130). I think it is obvious that we sometimes use self-ascriptions expressively, and in saying so I thereby add another expressive self-ascription to the tally. But I have now argued that only an expressivist account of self-ascriptions can provide a guarantee that truth and sincerity will coincide. If we do have first-person authority, then there is some such guarantee. A fortiori, if we have first-person authority, then we use expressive self-ascriptions at least as often as we make authoritative self-ascriptions. Quack erat demonstrandum.
