Précis of Dorit Bar-On’s Speaking My Mind: Expression and Self-Knowledge
- First Online:
- Cite this article as:
- Bar-On, D. Acta Anal (2010) 25: 1. doi:10.1007/s12136-009-0081-1
KeywordsAvowals’ security Immunity to error Moore’s Paradox Expression Showing Constitutivism
Consider everyday utterances such as ‘I’m feeling very tired’, ‘I’m scared of that dog’, ‘I’m wondering whether it’s going to rain’. Like other ordinary pronouncements, such utterances—‘avowals’, as they are often called—appear to inform us of certain contingent states of affairs. Semantically speaking, an avowal will typically identify an individual—the speaker—and ascribe to her an occurrent mental state—feeling thirsty, feeling scared of the dog, hoping that it doesn’t rain, etc. It will be true in the same circumstances as any ascription that identified that same individual and ascribed to her the same condition at the same time (‘She/DB is/was feeling very tired’). It can also serve as a premise in humdrum logical inferences (‘I feel nervous, and so do you; so that makes two of us’). I refer to the claim that avowals are truth-evaluable, and share grammatical and logico-semantic structure with other ascriptions, as Semantic Continuity.
However, epistemically speaking, avowals seem quite different from other empirical pronouncements, including ascriptions of mental states to others (‘She has a headache’), past-tense mental self-ascriptions (‘I felt awful last night’), self-ascriptions of psychological traits (‘I’m a very patient person’), as well as simple perceptual reports (‘There’s a bird on the feeder’), perceptual self-reports (‘I’m hearing a loud noise’) and proprioceptive or kinesthetic reports of bodily states (‘My legs are crossed’, ‘I’m sitting down’). Although avowals appear not to be made on any ordinary evidential basis, they are not normally subjected to ordinary kinds of epistemic assessment. At the same time, there is normally a strong presumption that a person’s avowals are true, and, except under unusual circumstances, an avowal is not to be rejected, denied, or criticized as unfounded. This is not just a matter of social decorum or the pragmatics of speech. If we consider, hypothetically, a ‘thought-avowal’—i.e., a present-tense self-ascription of a mental state spontaneously produced in thought only—we would also not presume to question or criticize it, wonder how the person knows it, or whether she has good reasons for it. For we presume that a person who avows, whether in speech or in thought, is in a privileged position to pronounce truly and knowledgeably on the subject-matter of her avowals—much more so than on any other topic. Moreover, although some perceptual reports and some nonmental self-reports can exhibit some first-person/third-person asymmetries, they are all open to what may be called brute error—an error that is simply due to the world failing to cooperate, rather than being due to some kind of failure of the subject’s conceptual, perceptual, or psychological faculties.1 By contrast, though we can perhaps envisage cases of false avowals—wishful thinking, self-deception, and so on—these are precisely cases in which the subject is thought to suffer some psychological irregularity, defect, or deficiency. None of the stock examples of false avowals present us with a subject who has simply been fooled into issuing a false avowal by an uncooperative mental world.2 I refer to the claim that there are these contrasts between avowals and all other ascriptions as Epistemic Asymmetry.3
Why is it that avowals, understood as true or false ascriptions of contingent states to an individual, are so rarely questioned or corrected, are generally so resistant to ordinary epistemic assessments, and are so strongly presumed to be true?
Do avowals serve to articulate privileged self-knowledge? If so, what qualifies avowals as articles of knowledge at all, and what is the source of the privileged status of this knowledge?
Avowals aside, what allows us to possess privileged self-knowledge?
The early chapters of SMM are devoted to a critical discussion of a variety of attempts to account for avowals’ security. I begin with a proposal that attributes the special security of avowals to the way the pronoun ‘I’ (or its equivalents) is typically used to pick out the subject of an avowal. (After all, ascribing an occurrent thought or feeling or sensation to myself under the description ‘DB,’ or ‘so & so’s philosophy teacher’ will not, in the normal case, afford me the security characteristic of an avowal). It is argued in Chapter II that excessive focus on the way ‘I’ refers generates a false dilemma. Impressed by the security of avowals as ‘I’-ascriptions, we may feel compelled to find a peculiar kind of object—such as a Cartesian Ego—to which an avowal refers. Alternatively, convinced that there are no such objects, we may conclude that avowals are actually ‘subjectless’—that they involve no reference to individuals at all. Both these alternatives are problematic as accounts of the semantic functioning of ‘I,’ and each in its way involves compromising Semantic Continuity. What is worse, neither alternative results in an acceptable explanation of the secure status of avowals.
Now, it might be thought that an easy way to avoid the dilemma is provided by the contemporary materialist incarnation of the Cartesian view. According to materialist introspectionism, the security of avowals is due to the operation of a special mechanism for detecting or ‘tracking’ our own mental states, which are, in turn, a subset of our internal physical states. As argued in a later chapter (Chapter IV), however, materialist introspectionism fails to account for Epistemic Asymmetry in its full scope, not the least because it implies that avowals should be straightforwardly open to brute error (just like proprioceptive reports, for example). I argue that this deficiency is shared by all other extant versions of what I call the ‘epistemic approach’, regardless of the particular epistemic route, method, or basis whose security they invoke to explain avowals’ distinctive security.
The seeds of a non-epistemic approach to avowals’ security are already sown in Chapter III, where I introduce an account of secure ‘I’-ascriptions that I take to be more promising than either the Cartesian or the ‘no reference’ accounts mentioned earlier. On this account, due to Sydney Shoemaker (1968) and Gareth Evans (1982), avowals belong in a class of ‘I’-ascriptions that enjoy a distinctive epistemic security—what Shoemaker has dubbed ‘immunity to error through misidentification.’ Thus, briefly, consider self-ascriptions such as ‘My legs are crossed’, ‘I’m sitting down’, or ‘I see a canary’6. Semantically speaking, these are predications of specific properties to a particular individual, myself. But, as typically made, such self-ascriptions contrast epistemically with analogous third-person ascriptions (e.g., ‘Her legs are crossed,’ or ‘She’s sitting down’) in that they are not grounded in a separate identification of some individual as oneself coupled with the recognition that that individual has the relevant property. When making a self-ascription of this sort, a subject is immune to a specific kind of error—an error due to misidentifying the individual who is F. As Evans observes, when a self-ascription of the form ‘I am F’ is immune to error through misidentification, although I may fail to be F, so my self-ascription may be false, there is no room for me to wonder: Someone is F, but is it me? This is because in such cases I have no grounds for thinking that someone has the relevant properties over and above, or separately from, any grounds I might have for thinking that I have them. But, importantly, the immunity in question is a sort of negative immunity to a certain kind of failure; it is not a reflection of positive recognitional success—success in recognizing the ‘right’ individual of whom to predicate F. When I say or think in the ordinary way: ‘I see a canary,’ it is not as though I think of someone in particular that she sees a canary, and correctly judge that it’s me who sees it. It’s for this reason that, if I am wrong about my seeing a canary (which, of course, I can be), it will not be because I failed to identify myself as the correct subject of my self-ascription.
Though not absolutely infallible, they are immune to a certain kind or error and doubt (concerning the identity of the subject of ascription).
They represent semantically structured ascriptions that genuinely refer to individuals and ascribe to them properties.
They do so without relying on a positive recognitional identification (i.e., the judgment that I am F does not rest on a recognitional identification of myself).
I propose (in Chapter VI) that we can apply this model to the ascriptive component of avowals.7 Specifically, I suggest that although, semantically speaking, when avowing, we issue genuine self-ascriptions that can be true or false, epistemically speaking, our avowals enjoy not only immunity to error through misidentification but also immunity to error through misascription (where the latter notion is explained by analogy to the former). The key idea is as follows. When avowing (as opposed to issuing a mental ascription, to oneself or to others, on the basis of observation, evidence, inference, analysis, or interpretation), a subject has no more reason, or epistemic grounds, for affirming components of the self-ascription than she has for issuing the self-ascription as a whole. When I avow being scared of that dog, say, I have no more reason for thinking that someone (though perhaps not me) is scared of that dog, or that I’m feeling something about that dog, (though perhaps not feeling scared), or that I’m feeling scared of something (though perhaps not of that dog), than whatever reason I have for thinking, simply, that I’m scared of that dog.8
The characterization of avowals as enjoying a distinctive ascriptive immunity to error provides a suitably tempered take on the claim that avowals are infallible, incorrigible and indubitable, which does not require invoking either Cartesian privileged access or, indeed, any distinctively secure epistemic basis on which avowals are said to rest. The characterization also gives a more precise sense to the intuition that avowals are epistemically simpler and more immediate or direct than other ascriptions, despite being semantically continuous with them. But, if we are to have a satisfactory account of Epistemic Asymmetry, we still need to identify the source of the ascriptive immunity of avowals. Toward this end, I consider a special case of avowals: self-verifying avowals of presently entertained thoughts (e.g., ‘I’m thinking that it’s time to go’). I argue that this case exhibits some key features that are shared by all avowals. Using an insight that is derived from Chapter V’s discussion of content skepticism and content externalism, I locate the source of avowals’ distinctive security—what accounts for their immunity to error of misascription—in their character as expressive acts in which a subject shares, gives voice to, or just vents, a self-ascribed mental state.
My positive neo-expressivist proposal is that the epistemic simplicity and immediacy of avowals can be explained by their being expressions of subjects’ self-ascribed mental states. Chapters VII and VIII of SMM are devoted to developing in detail this expressivist insight, which I borrow from the oft-dismissed traditional ‘simple’ avowal expressivism that many find in Wittgenstein. The core expressivist claim I incorporate into my account is that avowals—whether phenomenal (‘I feel so achy’) or intentional (‘I’m scared of that dog!’, ‘I am wondering whether he’ll show up at the party’)—can be seen as pieces of expressive behavior, similar in certain ways to bits of behavior that naturally express subjects’ states. Like simple avowal expressivism, neo-expressivism regards avowals’ distinctive security as due to their expressive character, rather than being inherited from the security of this or that epistemic basis on which they are made. It seizes on the epistemic similarities between avowals and other acts in which we give expression to our present states of mind (saying ‘What a mess!’, or ‘Ugh!’, or making an exasperated gesture, upon seeing a child’s messy room). But in contrast to simple avowal expressivism, neo-expressivism recognizes important dissimilarities between avowals and inarticulate grunts, grimaces, or cries, and even many verbal expressions. For like various mental and non-mental descriptive reports, avowals have products—sentence- or thought- tokens—that are semantically complex and are truth-evaluable. When avowing, one acts so as to articulate, out loud or silently, a present state of mind. When performed out loud, avowals are acts of speaking one’s mind self-ascriptively, in lieu of giving either non-linguistic or else non-self-ascriptive expression to one’s state of mind. Understanding the similarities between avowals and these other sorts of expressive behavior is crucial to providing an alternative to the epistemic approach, whereas understanding the differences is crucial to respecting Semantic Continuity (and thereby avoiding key mistakes of simple avowal expressivism).
According to neo-expressivism, if a person’s avowals are not to be questioned, corrected, second-guessed, expected to be supported by evidence, and so on, this is because—or to the extent that –they are taken to express the avowed states of mind, much like spontaneous natural expressions. But unlike grimaces we produce when in pain, or grunts we produce when displeased, the avowal tokens we produce can be true or false. Why then do we so strongly presume them to be true? The answer is that these tokens are self-ascriptions of states of minds that the acts of producing them serve to express. When avowing, I act so as to express the very same state that the avowal understood as product—i.e., the sentence-(or thought-) token I produce—says I am in. But then to take me to be avowing is to take it that I am in the very state whose presence is required to make my avowal true.
Neo-expressivism can thus explain the strong presumption of truth that governs avowals, as well as their distinctive epistemic ‘profile’. Like so-called constitutivist views, and in contrast to epistemic views, the neo-expressivist view explains avowals’ distinctive features without appeal to some secure epistemic basis on which they are made. Instead, like simple avowal expressivism, it appeals to avowals’ expressive character. However, unlike simple expressivism, and unlike some constitutivist views, it does not take Epistemic Asymmetry to be built into the semantics or ‘logical grammar’ of mentalistic discourse; and it does not require denying that avowals, qua products, are genuine self-ascriptions that can share truth-conditions with other ascriptions. Thus, it respects Semantic Continuity.9
The neo-expressivism developed in SMM also improves considerably on traditional avowal expressivism by providing an account of the sense in which avowals express states of mind. But, by itself, the account of avowals’ security it offers does not explain the special claim to knowledge subjects have with respect to their present states of mind, and it does not provide an explanation of the special status of self-knowledge. Indeed, the account is compatible with a deflationary view, which maintains that the seemingly privileged status of self-knowledge is fully exhausted by the special security of avowals understood non-epistemically. However, in Chapter IX I argue that the deflationary view is not entailed by neo-expressivism. Since I believe subjects do have a special claim to knowledge of their own present mind, I am concerned to show that my neo-expressivist account is at least compatible with a substantive answer to question (ii) above, and I argue that the account of avowals I offer meets this desideratum. I sketch several proposals concerning the privileged status of self-knowledge that are compatible with my account of avowals. I also offer a more direct explanation of the privileged status of self-knowledge in terms of the notion of ‘first-person privilege’ (as contrasted with so-called ‘first-person authority’).
In SMM’s final Chapter, X, I present some reflections on possible implications of the neo-expressivist account of avowals for certain metaphysical issues in the philosophy of mind. In particular, the chapter addresses two metaphysical issues that are kept in the background earlier: the reality of mental states, and the extent to which they are ontologically independent from expressive behavior. The chapter concludes by explaining how neo-expressivism meets all the desiderata laid down in the opening chapter. In addition to accommodating Epistemic Asymmetry in its full scope while respecting Semantic Continuity, an adequate account should, I claim, avoid committing to the Cartesian view of avowals as absolutely infallible or incorrigible (in keeping with the commonsense treatment of avowals), as well as avoid committing to a deflationary view of self-knowledge and to Cartesian ontology. Although various accounts of avowals’ security currently on offer succeed to varying degrees in meeting this or that desideratum, I argue that none meets all the above desiderata in a satisfactory way. I hope readers of SMM come away persuaded that the neo-expressivist account does better.
It is for this reason perhaps that full-blown ‘internal world skepticism’ seems even less coherent that Cartesian wholesale external world skepticism. For discussion of the differences between these two sorts of skepticism, see Bar-On (forthcoming).
The foregoing is offered in SMM’s introductory chapter only as a first-pass characterization of Epistemic Asymmetry; my preferred characterization, partly in terms of the notion of immunity to error, is not fully developed until later in the book.
For discussion of how leading accounts currently on offer fail to respect one or the other claim, see Bar-On and Long (2001). Smith (1998) has independently argued that the main difficulty in reaching a satisfactory philosophical account of so-called first-person authority is to reconcile (what I here call) Epistemic Asymmetry and Semantic Continuity. Focusing on the difficulties in accounting for Epistemic Asymmetry, Wright et al. (1998) also points out the importance of preserving features grouped here under Semantic Continuity.
The Cartesian tradition also endorses self-intimation: the idea that if a subject is in a mental state then she is somehow guaranteed to know it, an idea that I set aside in my investigation.
Understood as a perceptual self-report, rather than as an avowal of a visual sensation.
The ground for this proposal is prepared in Chapter V, where I discuss a familiar argument that, if content externalism is true, our ordinary avowals of contentful mental states (e.g., ‘I’m thinking that water is wet’) cannot represent secure (let alone) privileged knowledge. I argue in that chapter that a crucial step toward rebutting this argument is to see that such avowals do not articulate recognitional self-judgments.
As I sometimes put it, my self-ascription does not rest, epistemically, on a ‘judgment of appearances’ regarding the bearer of the self-ascribed state, the presence or character of the state, or its intentional content. If my self-ascription is false (which can happen on my view), this will not be because I was misled by the way things appeared to me ‘as I looked inside’. Just as, on Evans/Shoemaker’s analysis, a proprioceptive report ‘I am F’ (e.g., ‘I am raising my arm’) is not epistemically grounded in ‘someone is F’ and ‘I am that someone,’ so (I maintain), in addition, an avowal ‘I am in M (with content c)’ is not epistemically grounded in ‘I am in some state (with this or that intentional content)’ and ‘That state is M (and its content is c)’.