1 Introduction: The Social Dimension of First-Person Authority

It is an essential part of our ordinary conception of self-knowledge that each person enjoys a distinctive epistemic security in ascribing occurrent mental states to the self. The distinctive feature of self-knowledge, which Davidson (1984) calls first-person authority, is reflected in our pre-theoretical observation that one cannot be usually mistaken about what one is thinking about right now. Whenever one sincerely ascribes an occurrent thought to oneself by uttering, e.g., “I believe that Wagner died happily,” there is normally a strong presumption of truth: if the speaker sincerely says she is entertaining a certain thought right now, then she usually (or always) has it. Correlatively, one’s interlocutor is supposed to take the truth of the self-ascription for granted; and, under normal circumstances, it is even unclear how she might reasonably challenge it. Self-ascriptions of mental states are also characteristically direct: one can simply know one’s own mind without relying on any evidence, and “How do you know?” is usually a non-sensical question to be raised against sincere mental self-ascriptions. One’s ascriptions of mental states to others, on the other hand, do not share any of these features. They need to be based on evidence, they can often turn out to be false, and they can be easily challenged. Mental ascriptions to the self and to others thus seem to exhibit some distinctive kind of asymmetry. Many have attempted to explain it, including Davidson himself.^1,2

As some authors recently point out, it is important to notice that the notion of first-person authority has an importantly social dimension (Borgoni 2019; Winokur 2023; Tanaka 2024; see also Schwengerer 2021 for a similar discussion). It contains not just the idea that one’s sincere mental self-ascriptions are usually (or always) true despite the absence of evidence, but also the idea that one’s interlocutor is supposed to treat such mental self-ascriptions as true and can thereby gain knowledge about the speaker’s mind by simply accepting them. These two aspects of first-person authority are closely related but conceptually distinct. The following scenario provided by Borgoni (2019), for example, demonstrates this point vividly by showing how they could in principle come apart:

Imagine yourself being a woman (or someone belonging to a gender, race, or other target group of prejudice). You tell your colleague, after having considered a candidate’s application portfolio for an open position in your department, “I disagree with your proposal. I believe that this is the person we should hire.” Your colleague, dismissing what you have just said and appealing to prejudicial ideas replies, “No, you don’t believe it. Women just like to fuss with men’s decisions” (Borgoni 2019, 296).³

In this case, although the woman’s mental self-ascription, “I believe that this is the person we should hire,” is true, her colleague does not take it as true (due to a sexist prejudice, in this case). As Borgoni puts it, first-person authority is “not granted unless our peers recognize it in our words” (2019, 298, emphasis mine).⁴

Apparently, however, the social dimension of first-person authority has not been the primary focus of the discussion in the literature. See, for example, the formulation of the notion of first-person authority found in Coliva (2016):

Given C-conditions (including concepts’ possession, cognitive well-functioning, alertness and attentiveness), if one judges to have a mental state M (save for dispositional ones or for the dispositional elements of some mental states), one will usually [or, always, in her stronger formulation] have it. (Coliva 2016, 65)

The formulation captures the idea that one’s mental self-ascription is always or usually true despite the absence of evidence, but it does not explicitly introduce the perspective of the interlocutor. It does not mention the idea that one’s interlocutor is supposed to take such mental self-ascriptions to be true and can thereby gain knowledge about the speaker’s mind. Both may usually go together, but they are nonetheless distinct and can come apart in some cases. The former concerns a speaker’s ability to make true mental self-ascriptions, and as such, it captures the internal, or personal, dimension of first-person authority. The latter concerns an interlocutor’s ability to gain knowledge about the speaker’s mind by attaching the presumption of truth to her mental self-ascriptions, and as such, it captures the relational, or social, dimension of first-person authority.

A philosophical account of first-person authority should be able to incorporate both internal and relational aspects of the phenomenon. More specifically, one’s account of first-person authority should be able to explain how a speaker’s mental self-ascriptions can be usually (or always) true despite the absence of evidence, in such a way that it also explains, or at least does not make unintelligible, how one’s interlocutor could acquire knowledge about the speaker’s states of mind often enough by taking such mental self-ascriptions at face value (Tanaka 2024). With this in mind, it would be an interesting project to examine various accounts of first-person authority developed in the past few decades by asking whether it can meet this theoretical constraint that has started to attract some attention in the recent discussions.

In this paper, I will reexamine Davidson’s account of first-person authority in light of this recent discussion and argue that it has interesting theoretical resources to address both the personal and the social aspects of first-person authority within a unified framework. Below, I will roughly explain the outline of the main discussion developed in the subsequent sections.

Among commentators on Davidson, there is a consensus that Davidson’s official account of first-person authority invokes two key ideas. The first is the idea that first-person authority can be explained in terms of a speaker’s authoritative knowledge of the meanings of her own words. The second is the idea that the special status of a speaker’s semantic knowledge is to be explained in terms of a non-communitarian externalist view of meaning-determination. According to this view, roughly, the meanings of the words produced on the lips of an individual are generally determined by how that individual herself is disposed to apply them to objects in the environment. A speaker cannot be wrong about what her words deployed in mental self-ascriptive utterances mean, because it is the speaker’s own use that confers meaning to those words in the first place.

As I will argue, to see a fuller picture of Davidson’s view on first-person authority, it needs to be integrated with his discussion on triangulation, where he purports to show that meaning should be essentially social and interpretable. My strategy for demonstrating this point will be to formulate an apparent puzzle for Davidson concerning the social dimension of first-person authority and explain how it can be resolved in his theoretical framework. Davidson’s claim that meaning is determined in the non-communitarian way, when taken in isolation, may seem to suggest that there is no external social constraint on what a speaker can mean with her linguistic expressions. From this, one could be misled to think that Davidson’s account will actually face the skeptical worry concerning knowledge about other minds: the contents of the beliefs that a speaker expresses by making self-ascriptive utterances could be absolutely private in the sense that no person other than the speaker herself can ever correctly articulate them. If so, one might think that even if Davidson succeeds in explaining how a speaker’s mental self-ascriptions cannot be usually mistaken, he does so at the cost of leaving it mysterious how a hearer can attach the presumption of truth to the speaker’s mental self-ascriptions and thereby obtain knowledge about the speaker’s mind. In short, the worry is whether Davidson’s account of first-person authority can successfully capture the social dimension of first-person authority. To resolve this problem, we will need to look at his discussion of triangulation. As we will see, Davidson’s view is that a speaker could not mean anything by a word unless it is in principle interpretable by another person. This, by itself, may be an idea familiar to the readers of Davidson. However, Davidson does not elaborate on the social aspect of his account of meaning in the papers where he discusses the issue of first-person authority, and commentators have not examined exactly how these two issues are connected in Davidson’s philosophical system, taken as a whole. The aim of this paper is to fill this gap in the literature and show how Davidson’s view, when examined holistically, is indeed tailored to capturing the social dimension of first-person authority.

The discussion proceeds as follows. In Section 2, I will explain the two key ideas that constitute Davidson’s official account of first-person authority more carefully. In Section 3, I will formulate the skeptical worry concerning knowledge of other minds that might seem to follow from Davidson’s non-communitarian view of meaning-determination. In Section 4, drawing on his discussion of triangulation, I will explain how Davidson’s view of meaning-determination is crucially social in its own way and argue that it has an important implication on the issue of interpretability. In Section 5, I will connect Davidson’s view on the social nature of meaning with his account of first-person authority to demonstrate how this allows Davidson to dissolve the skeptical worry and thereby incorporate the social dimension of first-person authority. Section 6 is the conclusion.

2 Davidson’s Account of First-Person Authority

2.1 First-person authority explained by speaker’s authoritative knowledge of meaning

There is a consensus among commentators that Davidson’s basic explanatory strategy as developed in Davidson (1984) and (1987) (“First-person Authority” and “Knowing One’s Own Mind”) is to show how we can understand the phenomenon of first-person authority concerning mental self-ascriptions in terms of the authoritative knowledge of meaning that a speaker possesses with regard to her self-ascriptive utterances, such as “I believe that Wagner died happily”.⁵ This is the first key component in Davidson’s account of first-person authority. Here is how he sets up the discussion: to know what belief a speaker expressed by a certain utterance, in general, one should know the meaning of the sentence uttered, as well as whether the speaker takes the sentence to be true. This is how Davidson himself puts it: “if you or I or anyone knows that I hold this sentence true on this occasion of utterance, and she knows what I meant by this sentence on this occasion of utterance, then she knows what I believe—what belief I expressed” (Davidson 1984, 11). Davidson, for the sake of discussion, simply assumes that both the speaker and the hearer know that the speaker takes the sentence to be true. Then, he proposes to explore whether first-person authority could be explained by the idea that the speaker can know the meanings of one’s own words in some distinctively secure way that is not available to the hearer.

Davidson proposes that first-person authority derives “from the fact that the assumption that I know what I mean necessarily gives me, but not you, knowledge of what belief I expressed by my utterance” (1984, 12). More precisely, Davidson points out that we should look at a distinctive asymmetry that holds between a speaker’s knowledge of the meaning of her own self-ascriptive utterance and a hearer’s knowledge about it: there is “a presumption that speakers, but not their interpreters, are not wrong about what their words mean” (1984, 12). Davidson notes that this is not to say that speakers are never wrong about the meanings of their words (1984, 12). Rather, the relevant asymmetry is that speakers cannot be regularly and systematically mistaken about the meanings of their words, whereas that is perfectly possible for their interlocutors. This, on Davidson’s view, is the source of first-person authority: when a speaker ascribes a belief to herself by making a self-ascriptive utterance, “I believe that Wagner died happily,” the speaker knows the meaning of “Wagner died happily” as uttered by herself in some distinctively secure way that is not available to the hearer. So, Davidson claims, the speaker knows what belief is ascribed to herself by the utterance in question, in some distinctively secure way that is not available to the hearer. This is how he summarizes his proposal: “[there is] a presumption…that the speaker usually knows what he means. So there is a presumption that if he knows that he holds a sentence true, he knows what he believes” (1984, 14).

To complete his account, however, Davidson needs to explain the source of the presumption that a speaker is not regularly mistaken about the meanings of her own words. Without some such explanation, it would seem that Davidson’s account simply replaces one philosophical problem (i.e., first-person authority concerning belief self-attributions) with another (i.e., a speaker’s authoritative knowledge of her own meaning). In Davidson (1984), he states that the asymmetry in question is reflected in the idea that a speaker can know the meaning of her own utterance directly, whereas the hearer must rely on evidence.⁶ This, however, seems to provide a mere description of the relevant asymmetry, rather than an explanation.⁷ Commentators point out that Davidson’s explanation of a speaker’s authoritative knowledge of meaning—the second key component in his overall account of first-person authority—can be found in his later writings on this topic.

2.2 Speaker’s authoritative semantic knowledge explained by non-communitarian semantic externalism

Here is a key passage from Davidson (1987):

The explanation [of first person authority] comes with the realization that what a person’s words mean depends in the most basic cases on the kinds of objects and events that have caused the person to hold the words to be applicable…An interpreter of another’s words…must depend on scattered information, fortunate training, and imaginative surmise in coming to understand the other. The agent herself however, is not in a position to wonder whether she is generally using her own words to apply to the right objects and events, since whatever she regularly does apply them to gives her words the meaning they have and her thoughts the contents they have…To put the matter another way, nothing could count as someone regularly misapplying her own word. (Davidson 1987, 37–38)

The passage contains two important claims that need to be unpacked. Let me explain each in turn.

As Child (2007) helpfully observes, for Davidson, “the fundamental justification for the presumption that speakers generally know what their words mean comes from considerations about meaning-determination” (Child 2007, 165). Importantly, what is at work here is the view that we might call non-communitarian externalism, according to which the meanings of words uttered on the lips of a speaker are determined generally by what objects and events that the speaker regularly applies the words to. This is a form of externalism, because the meanings of words in a speaker’s language are determined by causal interactions between the speaker and objects/events external to her.⁸ Furthermore, the view is non-communitarian in that the meanings of words in a certain speaker’s language are determined generally by that speaker’s patterns of language use. That is, Davidson’s externalism is crucially different from the Burgean communitarian externalist view, according to which the meanings of one’s words are determined in part by how other speakers in the same linguistic community tend to use them (Burge 1979).

So, the first key idea presented in the above-cited passage is this: a speaker cannot make regular errors in using her own words, because their meanings are determined generally by the speaker’s overall patterns of use in the first place. To use an analogy by Child: a person cannot regularly act out of character because the person’s character is determined by how she generally acts; similarly, a speaker cannot regularly misuse her words because their meanings are determined by how the speaker generally uses them (Child 2013, 539). The hearer’s usage of the same-sounding words, on the other hand, does not similarly contribute to the meaning-determination of the speaker’s words. Davidson thinks that this asymmetry in meaning-determination explains the distinctive security of a speaker’s knowledge of the meanings of one’s own words.

Indeed, Davidson contends that those who adopt the Burgean communitarian externalist view would have a serious problem in explaining first-person authority. As is well-known, the Burgean view allows for the possibility that a competent speaker can be radically mistaken about the meaning of a term in her vocabulary and regularly use it in a confused way (e.g., “I have arthritis in my thigh,” where “arthritis” means joint inflammation). Davidson explicitly claims that this would undermine a speaker’s authoritative knowledge of her own meaning (Davidson 1987, 26–27).⁹ Reflecting this, in a later article, Davidson notes that he endorses “a fairly extreme form of individualism about meaning,” which is “not accepted by those who think languages are defined by shared practices” (Davidson 1993a, 250). On Davidson’s view of meaning-determination, the same-sounding words produced on the lips of different speakers may, and often do, have different meanings in their idiolects. Of course, they might sometimes converge with one another to some extent. But Davidson’s point is that they need not.¹⁰

Now, let us move on to the second key point. Construed in this way, there seems to be an immediate objection to Davidson’s explanation: even if a speaker’s words in her language are determined generally by the speaker’s own use, it does not show that the speaker is thereby in a privileged position to know what their meanings are. To use Child’s analogy again: even if a person character is determined by how she generally acts, she may not necessarily know what her character is—it requires observing her actual behaviors in various circumstances (Child 2013, 539). One way to respond to this objection on behalf of Davidson is to think that he endorses what we may call the deflationary conception of knowledge of meaning. On the deflationary conception, a speaker’s knowledge of the meanings of her own words can be identified with the speaker’s ability (knowing-how) to use them correctly, instead of the speaker’s possession of some propositional knowledge about their meanings.¹¹ This is to think that Davidson understands a speaker’s authoritative knowledge of the meanings of her own words in terms of the speaker’s ability to use them in a way that is immune to regular errors. As he puts it in the passage cited above, “nothing could count as someone regularly misapplying her own word”.¹² Arguably, this kind of linguistic/semantic knowledge-how that constitutes a speaker’s linguistic competence need not consist in the propositional knowledge that one’s words mean such and such.^13,14 If this is the conception of semantic knowledge that Davidson needs, then the objection can be avoided. (To clarify, some commentators such as Child argue that we could also find a richer, non-deflationary conception of semantic knowledge in Davidson. Although the plausibility of the discussion below will not hang on this issue, see footnote for discussion).¹⁵

To see how this allows Davidson to bypass the objection, it should be helpful to notice how it explains the required kind of directness of a speaker’s authoritative knowledge of meaning. A speaker’s authoritative knowledge of meaning (construed in a deflationary way) need not be supported by evidence, Davidson can say, because the knowledge involves no proposition to be supported in the first place. On the deflationary reading, roughly, being a competent user of a language suffices for knowing one’s own meanings. Just as one can know how to ride a bicycle without possessing any articulate propositional knowledge that necessary maneuvers are such and such, one can know how to correctly use words without possessing any propositional knowledge that their meanings are such and such. Indeed, Davidson seems to think that a fully competent speaker need not be able to articulate the meanings of her own words, other than in the purely disquotational form: “[t]he speaker, after bending whatever knowledge and craft he can to the task of saying what his words mean, cannot improve on the following sort of statement: ‘My utterance of ’Wagner died happy’ is true if and only if Wagner died happy’” (Davidson 1984, 13).

Let me summarize the overall structure of Davidson’s official account of first-person authority. The explanandum is the asymmetry that exists between a speaker and a hearer: by uttering a self-ascriptive utterance, the speaker can ascribe a belief to herself in the distinctively secure way that is not available to the hearer. Davidson thinks that the asymmetry can be explained in terms of the speaker’s knowing the meanings of her own words in the distinctively secure way that is not available to the hearer. On Davidson’s view, the asymmetry in semantic knowledge can be explained by his non-communitarian externalist view of meaning-determination: the speaker cannot regularly misuse her own words because it is the speaker’s regular use that determines their meanings, whereas a hearer does not similarly contribute to the determination of their meanings. Lastly, in order for Davidson’s account to work, apparently, he needs to adopt a deflationary conception of speaker’s knowledge of meaning.

In the rest of the paper, I will assume that this provides a by and large accurate reconstruction of Davidson’s official account of first-person authority.¹⁶ Below, I will argue that it is crucial to connect Davidson’s account of first-person authority with his fuller account of meaning-determination that draws on the triangulation argument, because failing to do so can mislead us to have the impression that due to his commitment to “a fairly extreme form of individualism about meaning,” Davidson in fact fails to incorporate the social dimension of first-person authority in his framework. In the next section, I will explain this point more carefully.

3 Skeptical Worry for Davidson?: Sharing Beliefs through Self-ascriptions

For the sake of discussion, I shall raise an objection to Davidson’s account of first-person authority along the following lines. His non-communitarian externalism, one might say, will prevent him from accommodating the social dimension of first-person authority we observed in Section 1. This is because the view in tension with what we may call the ordinary view concerning a hearer’s acceptance of a speaker’s self-ascriptive utterance. To introduce the ordinary view, consider this simple question: for a speaker, what is the point (or, purpose) of making a self-ascriptive utterance, after all? The point, at least in part, consists in allowing a hearer to acquire knowledge about the speaker’s current mental state through her comprehension of the speaker’s self-ascriptive utterance. Linguistically ascribing a mental state to the self is a way of sharing, as it were, the self-ascribed mental state to a linguistically competent hearer.¹⁷ As observed in Section 1, a certain methodological constraint that any adequate account of first-person authority should aim to meet: one’s account should explain first-person authority concerning a speaker’s mental self-ascriptions, in such a way that it also explains, or at least does not make unintelligible, how a hearer can attach the presumption of truth to the speaker’s mental self-ascriptions and thereby gain knowledge about the speaker’s mind by accepting them at face value.¹⁸

Those who find the ordinary view plausible would describe what is typically involved in a speaker’s self-ascriptive utterance and a hearer’s acquisition of knowledge about the speaker’s mind in the following way. Suppose that a speaker, Ann, ascribes a belief to herself by uttering, “I believe that Wagner died happily.” Suppose also that Ben, her interlocutor, hears every single word of her utterance correctly, and all of the words Ann uttered sound perfectly familiar and intelligible to his ears. Ben recognizes Ann’s first-person authority, and, assuming that Ann is sincere, Ben proceeds to take Ann’s words for granted. As a result, this should put Ben in a position to legitimately say, “You believe that Wagner died happily,” and acquire knowledge that Ann believes that Wagner died happily. There should be nothing puzzling here. After all, the purpose of Ann’s making the mental self-ascription is presumably to share her mental state with Ben, as highlighted in the recent discussion on the social dimension of first-person authority.¹⁹

Here, one may question whether Davidson can adequately explain the legitimacy of this ordinary practice in his framework. The question can be motivated by recognizing the fact that, in ascribing a belief to Ann, Ben should use his own words to specify the content of Ann’s belief. Apparently, if we adopt Davidson’s view of meaning-determination, Ben is not necessarily entitled to the presumption that he and Ann mean the same things by the same-sounding words, after all. Since Davidson rejects the Burgean communitarian externalism, Ann and Ben may be using the same-sounding words with different meanings in their own idiolects. If this is so, to acquire knowledge about Ann’s mental state, Ben would have to first confirm whether the meanings he attaches to the relevant words (i.e., “Wagner,” “died,” “happily”) match the meanings Ann attaches to the same-sounding words in her language.²⁰

So far, this by itself does not pose any serious problem yet. Occasional misunderstandings may be simply common. Indeed, Davidson’s point is precisely that there is such a speaker-hearer asymmetry: while a speaker cannot be generally mistaken about the meanings of her own words, a hearer can.²¹ However, by pressing this line of thought further, we will face an important question. In Davidson’s framework, is it guaranteed that Ben could in principle correctly specify the meanings of the relevant words in Ann’s language, and thereby correctly articulate the content of the belief Ann ascribes to herself? Or, is it possible that, no matter what evidence Ben manages to collect concerning Ann’s linguistic behaviors, Ben could never be able to correctly articulate the content of Ann’s belief, using his own words? The question concerns whether Davidson’s non-communitarian view of meaning-determination may prevent him from providing a plausible explanation of how a speaker can come to share her belief with a hearer via language—in other words, how a speaker can make available the content of her current belief to a speaker by making a self-ascriptive utterance.

The question could turn into a skeptical worry, if one thinks that in rejecting the Burgean communitarian externalism, Davidson subscribes to the view that a person’s having a history of applying symbols to objects and events in the environment in some regular ways is sufficient for the symbols to have determinate meanings. This, in effect, would imply that a speaker’s idiolect could emerge as a solitary language: a language that one develops only for herself from scratch without relying on any social interaction. If this is Davidson’s view, it seems that he leaves it unexplained how a hearer can come to acquire knowledge about the speaker’s current state of mind by taking the speaker’s self-ascriptive utterance as true. For, if expressions used in the speaker’s self-ascriptive utterances belong to her solitary language, it is unclear how others can come to use them with the meanings they have in the speaker’s language. In the worst case scenario, it might even seem that, for a hearer to understand the solitary language developed by the speaker, she would need to become the speaker herself—that is, for two people to use the same sounding words with the same meanings, they would have to share the exact same history of applying those words to objects and events in the environment. If so, it would become realistically impossible for any two speakers to share meaning. This, if true, would render entirely ungrounded what I described as the ordinary view concerning a hearer’s acceptance of a speaker’s self-ascriptive utterance.

If Davidson’s account of first-person authority is indeed susceptible to this kind of skeptical worry, this would be a serious flaw: even if it explains the personal dimension of first-person authority, it does so at the cost of making its social dimension completely unintelligible. Of course, as I will explain, Davidson has a plausible answer to this worry. This becomes clear once we recognize that Davidson in fact does not think that a person’s having a history of applying symbols to objects and events in the environment in some regular ways is sufficient for the symbols to have determinate meanings. However, crucially, if we focus exclusively on his remarks on the issue of meaning-determination offered in his discussion of first-person authority, we can be misled to think otherwise. In the key passage I cited in the previous section, Davidson claims that “what a person’s words mean depends in the most basic cases on the kinds of objects and events that have caused the person to hold the words to be applicable,” and “whatever she regularly does apply them to gives her words the meaning they have and her thoughts the contents they have” (Davidson 1987, 37). We have also seen that he describes his position as “a fairly extreme form of individualism about meaning” (Davidson 1993a, 250). From these remarks, one might take Davidson to be committed to the idea that a person’s having a history of applying symbols to objects and events in the environment in some regular ways is sufficient for the symbols to have determinate meanings.

Furthermore, some of the expository remarks on Davidson’s account of first-person authority made by commentators might also help generate the impression that Davidson’s view of meaning-determination is not only non-communitarian but thoroughly non-social. See, for example, the following passages from Jacobsen (2009) and Child (2013):

…if, as Davidson believes, immunity to error can be assigned to individual speakers, then each solitary speaker’s uses of her words must be assumed to confer meaning on them. A speaker’s immunity to regular verbal error is then an expression of the idea that only her own uses of words can create the word-world patterns needed to constitute her utterances of those words as meaning what they do. (Jacobsen 2009, 256)

The main substance of the explanation comes from the idea that the meanings of a speaker’s words are determined by her own use of those words. That is what underpins the presumption that a speaker usually knows what her words mean. (Child 2013, 534)

These passages can help generate the impression that Davidson is talking about a solitary language in his discussion of meaning-determination. At least, in their expositions of Davidson’s account of first-person authority, neither Jacobsen nor Child connect it with Davidson’s fuller account of meaning-determination developed elsewhere. Yalowitz (1999), even more explicitly, claims that no explanatory role needs to be given to social interactions in Davidson’s account of first-person authority and claims that it must “be applicable wholesale to the solitary language user” (Yalowitz 1999, 111).²²

To clarify, these commentators, with the exception of Yalowitz, do not explicitly attribute the thoroughly non-social account of meaning-determination to Davidson. Nonetheless, it also has to be noted that they do not attempt to connect Davidson’s account of first-person authority with his discussion on the sociality of meaning either. As a result, I think their expositions can help generate the false impression that Davidson’s account of first-person authority is subject to the skeptical worry concerning knowledge about other minds that I formulated above and thereby fails to incorporate the social dimension of first-person authority.

In the rest of this paper, I will explain how Davidson’s account of first-person authority is not in fact susceptible to the skeptical worry sketched above. To see this, we need to understand how Davidson’s fuller account of meaning-determination incorporates a social element despite being non-communitarian. I turn to this task in the next section.

4 Davidson’s Social Externalism: Triangulation Argument

My aim in this section is to provide a fuller reconstruction of Davidson’s account of meaning-determination only to the extent that it is necessary to respond to the apparent problem I raised for Davidson in the previous section. The idea that meaning should be essentially social constitutes, in fact, an important strand in Davidson’s philosophy. In his earlier writings, he develops it with the notion of radical interpretation; in his later writings, he prefers to do it with the notion of triangulation. I do not intend to cover every aspect of Davidson’s view on meaning, nor do I aim to offer a completely novel interpretation of Davidson on this issue. Rather, my aim is to demonstrate how his view on the essential sociality of meaning and his account of first-person authority are connected in important places. This will allow us to see how his view, examined holistically, is in fact tailored to understanding the social dimension of first-person authority.

From the discussion in Section 2, we already know that Davidson’s account of meaning-determination is externalist: “what a person’s words mean depends in the most basic cases on the kinds of objects and events that have caused the person to hold the words to be applicable” (Davidson 1987, 37). Elsewhere, he calls this idea “perceptual externalism” (Davidson 1990, 198) and notes that he is strongly committed to it (see also Davidson 2001a, 2).²³ Also, as I explained, Davidson’s view is non-communitarian: the meanings of words in a speaker’s idiolect need not be affected by other speakers’ usage. Again, the question is whether Davidson thinks the existence of such causal connections between a person’s words and objects and events in the environment is sufficient for meaning-determination.

As Davidson himself writes in the introduction of Subjective, Objective, Intersubjective, another key component in Davidson’s externalism is the idea that “language is necessarily social,” in the sense that “to have thoughts, and so to mean anything in speaking, it is necessary to understand, and be understood by, a second person” (Davidson 2001b, xv). In “Second Person,” Davidson puts the idea in this way: “[T]here would be no saying what a speaker was talking or thinking about, no basis for claiming he could locate objects in an objective space and time, without interaction with a second person” (Davidson 1992, 121), and this shows that “one’s first language cannot be a private language, that is, a language understood by only one creature” (Davidson 1992, 121).²⁴ In other words, Davidson’s view is that establishing causal connections between a person’s words and objects and events in the environment by itself is not sufficient for meaning-determination. Or, more accurately, it is not sufficient for meaning-constitution, because Davidson thinks that, at least in basic cases, there would be nothing that a speaker could talk or think about in the first place unless the speaker interacts with, and is understood by, other people. So, although Davidson rejects the Burgean communitarian externalism, his account of meaning-determination is social in its own way.²⁵ Why does Davidson think that meaning-constitution/determination requires interpersonal interactions? To answer this question, we need to look at his discussion of triangulation. Below, I will mostly draw on Davidson (2001a), “Externalisms,” where we can hope to find the most elaborate and accessible version of his triangulation argument in comparison to his earlier writings.²⁶

Again, Davidson endorses the idea that the meanings of a speaker’s words are determined by the typical causes of their tokenings. However, Davidson now claims that this does not give us a complete account of meaning-determination, because such “typical causes” of the tokenings of words can be grouped in a numerous different ways. Here is how Davidson puts the problem:

The problem of relevant similarity comes up also in connection with the idea that the content of a perceptual belief depends on its common or usual cause. Even to approach the question, we must have some idea which causes are being gathered together… any set of causes whatsoever will have endless properties in common… (Davidson 2001a, 5)

Perhaps the best way to understand this problem is to formulate it via the Wittgensteinian observation that in order for a word to have a determinate meaning, some objective condition for its correct application should be established (Davidson 2001a, 3–4). If, for example, “cow” means cow on a speaker’s lips, this makes the speaker’s application of “cow” to an object correct (at least in basic cases) if and only if the object is in fact a cow; and, this simultaneously makes the speaker’s application of “cow” to a fake cow incorrect, however similar it might be to real cows. Fake cows might be indeed very similar to cows, but this similarity is not “relevant,” in the sense that they should not be categorized as “cow” on the basis of this similarity. As Davidson says, “[f]ake cows are in one way relevantly similar to real cows—that’s why we make mistakes. But in another way, fake cows aren’t at all like cows—they don’t fall under the concept cow” (2001a, 3–4) Now, the question is this: what establishes the fact that the speaker’s application of “cow” to a fake cow is incorrect, even if it looks highly similar to real cows for the speaker and she even has a disposition to apply “cow” in such a case? What is the “relevant similarity” that a group of objects, to which “cow” (as the speaker uses it) is correctly applicable, must exhibit? And, more importantly, how are we to answer this question? The task, in effect, is to explain what it takes for a person to possess any concept at all (2001a, 4).

In outline, Davidson’s proposal is that an adequate solution to the problem of “relevant similarity” requires triangulation: two people interacting with one another and with the world. In other words, he thinks that a solitary person who does not interact with another person cannot answer this problem.²⁷ Davidson notes that he is convinced by Wittgenstein (or, Kripke’s (1982) discussion inspired by Wittgenstein) that “only social interaction brings with it the space in which the concepts of error, and so of meaning and of thought, can be given application” (2001a, 4).²⁸ To secure the possibility of genuine error/misapplication, Davidson argues, whether a speaker’s reaction to an object is “relevantly similar” to her past reactions should not be a matter that the speaker alone can decide from the solipsist perspective. If this matter were entirely up to the speaker’s subjective judgement, the distinction between what seems to be an instance of correct application for the speaker and what is in fact an instance of correct application would collapse; and, the possibility of genuine error/misapplication would be lost.²⁹ On Davidson’s view, the distinction in question becomes available to a person only if she interacts with another person in the environment.

More specifically, Davidson explains how triangulation allows a person to recognize the possibility of genuine error/misapplication in the following way (2001a, 6). Imagine that there are two people, A and B, who can observe one another’s reactions to objects in the environment. Suppose that A regularly responds to cows by making the sound “cow”; now, we have another person B, who can observe that there are such correlations between “cow” uttered by A and a certain group of objects, which do look similar with one another to B herself (2001a, 6). Now, B is in a position to “notice occasions on which the responses of the other differs from the other’s usual responses” (2001a, 6). That is, B can recognize when A applies “cow” to an object that, from B’s perspective, A should not categorize as “cow,” given A’s past usage of the term. In this situation, “A’s responses are not taken by B to be the same as A’s earlier responses, while B’s responses are the same as B’s earlier responses” (2001a, 6).³⁰ From her own perspective, B can judge A’s particular application of “cow” to an object in the current situation to be incorrect, in the sense that it is not relevantly similar to A’s past reactions. Davidson’s main idea is that, to secure the possibility of genuine error, it is crucial that there be such a judgement that could diverge from A’s judgement about the correctness of her own application of the term. That is, even if A, from her own perspective, thinks that her application is in accordance with her past usage and thus is correct, it should be possible that there could be a genuine challenge to such a judgement. Such a challenge, on Davidson’s view, can only be issued by another person, B, who has been observing A’s past applications of “cow” to objects in the environment.

Davidson argues that this triangle allows A (and B, because A is also in a parallel position to do all of the above for B) to recognize the possibility of genuine error. This would “make the difference between the cases where the creatures act in unison, and the cases where they do not, available to the creatures themselves” (Davidson 2001a, 6). That is, A (as well as B) is now in a position to recognize that even if she thinks that the object in front of her should be categorized as “cow,”” it might be mistaken. Although it is her past usage of “cow” that constitutes what it means for her, it is not entirely up to her to decide whether her particular application of “cow” in a given situation is correct or incorrect—that is, whether it is “relevantly similar” to her past usage. Davidson takes this triangulation argument to show that establishing the possibility of genuine error—that is, distinguishing “between a mere disposition and rule following” (2001a, 7)—requires two people interacting with one another in the shared environment. As he puts it, “an interconnected triangle such as this…constitutes a necessary condition for the existence of conceptualization, thought, and language” (2001a, 7).

For the purposes of this paper, it is crucial to see that, if the triangulation argument is successful, it imposes a social constraint on what a speaker can mean by her own words: they need to be interpretable (and in basic cases, have been actually interpreted) by another person. On my reconstruction of Davidson’s triangulation argument, the key idea is that the question whether A’s current application of a word is consistent with her past usage can be genuinely raised only if there is another person, B, who can challenge A’s own judgement on this matter. This, in turn, requires that A and B could at least agree on what A’s word means—e.g., whether “cow” should be applied to this particular object of their joint attention, given A’s past usage. In other words, although the meaning that “cow” has for A is constituted by A’s own use of the symbol (and need not be affected by B’s use of the same symbol), there is a crucial social constraint that another person B should be able to grasp the condition for correct application that “cow” has for A (i.e., what “rule” A is following in applying “cow”), based on her observations of A’s regular applications of “cow” to objects in the environment. This is why “what a speaker means by what he says, and hence the thoughts that can be expressed in language, are not accidentally connected with what a competent interpreter can make of them (2001a, 11).³¹

To clarify, in this picture, there is no requirement that A and B should use their words with the same meanings. For example, there is nothing that prevents B from using the same sound, “cow,” in whatever way she likes, insofar as her own use is also interpretable by A. Of course, it would certainly facilitate communication if A and B use “cow” with the same meaning. But they need not. Thus, the triangulation argument is in line with Davidson’s rejection of the Burgean communitarian externalist view.

To summarize: as we have seen in Section 2, Davidson endorses the idea that the meaning of a speaker’s word is determined by her past applications of the word to objects and events in the environments (i.e., the typical causes of its tokenings). However, for the word to have any determinate meaning (i.e., to have a condition for correct application), the possibility of genuine error/misapplication should be made available to the speaker. Davidson thinks whether the speaker’s particular application of a word is relevantly similar to her past applications (i.e., whether the application is correct) should not be something that the speaker herself could answer from the solipsist perspective. There should be another person observing the speaker’s reactions to objects in the environment, who can judge their correctness from a different perspective than the speaker’s own. Because meaning, in basic cases, can only emerge through this kind of interpersonal triangulation, Davidson’s view of meaning-determination (or, constitution) is in an important way social, although not communitarian. From this view, it follows that there could not be such a thing as a non-shareable solitary language and whatever a speaker can come to mean by using a certain symbol should be interpretable by another person. As I will explain in Section 5, recognizing this consequence of Davidson’s fuller account of meaning-determination is key to dissolving the worry I raised for Davidson in Section 3.

5 Davidson on the Social Dimension of First-Person Authority

The discussion in this section proceeds in two steps. In the first subsection, I will argue that the discussion in Section 4 allows Davidson to respond to the worry formulated in Section 3. To give more support to this claim, in the second subsection, I will explain how it also allows us to interpret some of the key passages from “First-person authority” and “Knowing One’s Own Mind,” where Davidson clearly hints at, but does not really elaborate on, the social aspect of his account of meaning-determination.

5.1 Davidson’s response to the skeptical worry

The worry that I raised for Davidson’s official account in Section 3 was this: on Davidson’s account, a speaker’s immunity to regular linguistic errors is explained by his non-communitarian semantic externalism, according to which the meanings of words in a particular speaker’s language are determined generally by the speaker’s usage. I explained how this might seem to be in tension with the ordinary view that, for a speaker, the point of making a mental self-ascription should be to allow a hearer to acquire knowledge about the speaker’s current mental state by taking her self-ascriptive utterance at face value. Davidson’s acceptance of “a fairly extreme form of individualism about meaning,” (Davidson 1993a, 250) might suggest that he admits the possibility of non-sharable solitary language. This can make room for the possibility that a hearer, no matter what evidence she may manage to collect about the speaker’s language use, will never be able to articulate the content of a speaker’s belief correctly, using her own words. This in turn seems to make Davidson’s account of first-person authority subject to the skeptical worry concerning knowledge about other minds: even if Davidson’s account can explain why there is a presumption that a speaker’s self-ascriptive utterance is correct, it does so in a way that makes it unintelligible how this can also allow a hearer to acquire knowledge about the speaker’s current state of mind by taking the speaker’s self-ascriptive utterance at face value. If so, his view would be accused of failing to incorporate the social dimension of first-person authority.

Now, given the discussion in the previous section, we can understand why Davidson’s account is not susceptible to this kind of worry. Recall that the key component in Davidson’s account of first-person authority is the idea that a speaker cannot regularly misuse her own words because it is the speaker’s use that confers meanings on them in the first place. Also, Davidson rejects the Burgean communitarian externalist view in part because he thinks it compromises first-person authority. His account of meaning-determination is thus individualist: a speaker need not use words in the same way as others do. However, we can now see that this captures merely one side of his account of meaning-determination, because he also argues, thorough the triangulation argument, that there can be nothing that a speaker can mean that is not in principle interpretable by another. It is crucial for a word to have any determinate meaning that some person other than the speaker herself could grasp its condition for correct application.

This allows us to resolve the problem in Davidson’s framework. Although each speaker is, as Davidson argues, authoritative in determining what her words mean, there is also a crucial social constraint: what they can mean should be in principle interpretable to others. Therefore, although there is a crucial asymmetry between a speaker and a hearer concerning their knowledge about the meanings of the speaker’s words, the asymmetry should not be taken to imply that there could be some content that is absolutely private to the speaker in the sense that others could never be able to articulate what it is. The possibility of such absolute privacy is ruled out by Davidson’s triangulation argument that meaning can emerge only if there are two people interpreting one another’s words in the shared environment.

When Davidson’s account of first-person authority is properly integrated with his triangulation argument, we can see that his overall account is actually tailored to providing a plausible response to the skepticism concerning knowledge about other minds and thereby capturing the social dimension of first-person authority. Davidson, at least in one place, seems to note this point himself. The following passage comes after his brief discussion of triangulation in “Epistemology Externalized”:

Knowledge of one’s own mind is personal. But what individuates that state at the same time makes it accessible to others, for the state is individuated by causal interplay among three elements: the thinker, others with whom he communicates, and an objective world they know they share. (Davidson 1990, 203–4)

In this passage, he suggests that there should be no real tension between his account of first-person authority and the ordinary view that the content of any belief that a speaker attributes to the self should be accessible to others. He briefly notes that this connects with his discussion on triangulation, but does not go on to explain how. Above, I offered a more thorough exposition of this point on behalf of Davidson by making the relevant connections more explicit.

5.2 Some suggestive passages from “First-person Authority” and “Knowing One’s Own Mind”

As I noted earlier, even in “First-person authority” and “Knowing One’s Own Mind,” we can find some passages where Davidson clearly hints at the idea that meaning is essentially social in the way that I explained above. Here is one such passage from “Knowing One’s Own Mind”:

Though I reject Burge’s insistence that we are bound to give a person’s words the meaning they have in his linguistic community, and to interpret his propositional attitudes on the same basis, I think there is a somewhat different, but very important, sense in which social factors do control what a speaker can mean by his words. If a speaker wishes to be understood, he must intend his words to be interpreted in a certain way, and so must intend to provide his audience with the clues they need to arrive at the intended interpretation. […] It is the requirement of learnability, interpretability, that provides the irreducible social factor, and that shows why someone can’t mean something by his words that can’t be correctly deciphered by another. (Davidson 1987, 28)

This passage clearly presents the idea that meaning is essentially social, and it has something to do with interpretability. Towards the end of the paper, he also makes a similar note that “[f]irst person authority, the social character of language, and the external determinants of thought and meaning go naturally together” (Davidson 1987, 38). The problem, again, is that he does not go on to explain exactly how they go together.

The discussion in the previous section can also help us understand some of the passages that may seem to pose a problem for commentators such as Jacobsen or Child, who endorse the standard interpretation that Davidson’s account of first-person authority rests on his theory of meaning-determination. For example, in the last paragraph of “First-person Authority,” Davidson notes this:

There is a presumption—an unavoidable presumption built into the nature of interpretation—that the speaker usually knows what he means. So there is a presumption that if he knows that he holds a sentence true, he knows what he believes. (Davidson 1984, 14)

Here, Davidson notes that his account of first-person authority results from the “nature of interpretation,” which may seem to implicate that it does not come from his account of meaning-determination. In fact, there is at least one commentator who criticizes the Jacobsen-Child line of interpretation by accusing them of ignoring this passage (Hossein Khani 2021, 6.1). However, we can see why Davidson’s remark here need not pose any trouble for the standard interpretation. In his system, the issues of interpretation and meaning-determination are in fact inseparable. For this reason, if his account of first-person authority is based on his view of meaning-determination, it must be also based on his view on the nature of interpretation, at least indirectly.

In light of this, we can also note that it is perhaps misleading, although not mistaken, to claim that in Davidson’s framework, “the fundamental justification for the presumption that speakers generally know what their words mean comes from considerations about meaning-determination” and “it does not come directly from considerations about the process or procedure of interpretation” (Child 2007, 165). This is misleading because it can mask the crucial point that the social constraint of interpretability is also a key component in Davidson’s account of meaning-determination after all. As I explained, this point is not fully transparent in Davidson’s discussion of first-person authority. My claim has been that failing to recognize this point can lead to the mistaken idea that Davidson’s account of first-person authority fails to recognize the social dimension of first-person authority. Showing why this is not the case takes the kind of interpretative work that I have undertaken in this paper.

6 Conclusion

The aim of this paper has been to show how Davidson’s account of first-person authority, his rejection of the Burgean communitarian view of meaning-determination, and his view on the essentially social nature of meaning are connected in several important places. Specifically, if one is misled to think that Davidson subscribes to the view of meaning that allows for the possibility of solitary language, it would seem that his account of first-person authority fails to recognize the social dimension of first-person authority and even becomes susceptible to the skeptical worry concerning knowledge about other minds. To understand Davidson’s answer to the problem, we need to recognize how his rejection of the Burgean communitarian idea and his insistence on the essential sociality of meaning ultimately hang together. This also allows us to interpret some suggestive remarks from “First-person Authority” and “Knowing One’s Own Mind” where Davidson hints at, but does not go on to fully explain, connections between the issues of first-person authority and interpretability. As a result, we are now in a better position to see how Davidson’s theoretical system as a whole is indeed tailored to capturing the social dimension of first-person authority. First-person authority concerns not just how a speaker can usually make true mental self-ascriptions, but also how a hearer can accept them at face value and thereby gain knowledge about the speaker’s mind. Although only the first, personal, dimension has been the main focus of the discussion in the literature, in Davidson’s original discussion, we can find interesting theoretical resources to address both personal and social dimensions of the phenomenon in a unified framework. My aim has been to establish this point carefully to show that Davidson’s discussion is worth reexamination in light of the recent discussion that highlights the essentially social dimension of first-person authority.

Acknowledgements

This work was supported by the Japan Society for the Promotion of Science (JSPS KAKENHI 21J00685 and 22KJ0522). I am grateful to Dorit Bar-On, Lionel Shapiro, Ben Winokur and two anonymous reviewers for their valuable comments on earlier versions of this paper.