Self‐Knowledge

Few would say that we are infallible with respect to our own states of mind. Still, our ordinary ways of thinking represent people as having an especially intimate relationship to their own attitudes, sensations, emotions, and so on. This special intimacy is suggested inter alia by the default assumption that when sincerely made, present-tense mental state self-ascriptions – statements like “My feet are aching”, “I want to leave the party”, “I think Justin is bored” – will be true. Following familiar usage, I will call such statements ‘self-ascriptions’, and will label this feature of them their ‘first-person authority’. It is natural to understand first-person authority as bound up with the fact that we are in an especially good position to know about our own states of mind, with the idea being that sincere self-ascriptions express this knowledge. Call this the ‘epistemic approach’ to understanding first-person authority. But there is also a prominent line of thought which rejects the epistemic approach. Expressivists about self-ascriptions think that it fails to capture a second feature of these statements, what I will call (following Bar-On (2004, 6 ff.)) their ‘epistemic asymmetry’ with other kinds of assertion. Self-ascriptions appear to be epistemically unlike other kinds of assertion in that it is typically inappropriate to ask a self-ascriber for the reasons or evidence on the basis on which she knows about her own state of mind. Nor is it clear what epistemic method or process might underwrite one’s ability to self-ascribe mental states. In response to such considerations, Expressivists offer a non-epistemic explanation of first-person authority, in terms of the idea that self-ascriptions are ‘avowals’, utterances whose primary function is to express the state they self-ascribe, rather than a second-order judgment to the effect that one is in this state. In my view, Expressivists are right to take seriously self-ascriptions’ epistemic asymmetry with other forms of assertion, and right also that doing so requires understanding self-ascriptions as avowals. But they are wrong to think that doing so requires a non-epistemic approach to first-person authority, and also wrong to find such an approach unproblematic. After explaining how Expressivism is motivated by the need to reconcile self-ascriptions’ first-person authority with their epistemic asymmetry (section 2), I argue that we should reject Expressivism because it doesn’t adequately capture the epistemology of self-ascription (section 3). This discussion motivates developing a version of the epistemic approach, albeit one which like Expressivism, puts self-ascriptions’ epistemic asymmetry centre stage. I do so by showing how viewing self-ascriptions as avowals can itself explain why they manifest genuinely epistemically ungrounded self-knowledge (sections 4-5, 4-5). The account works by locating the phenomena of avowal, self-knowledge, and self-ascription within a ‘forward-looking’ epistemology, on which propositional knowledge is viewed as a certain kind of rational ability. Here I detail the epistemic approach to understanding first-person authority (2.1), and explain why it might seem unable to accommodate their epistemic asymmetry (2.2). I then describe the Expressivist’s non-epistemic response to these considerations (2.3). Contrast self-ascriptions with ascriptions of mental states to others. Seeing Justin across the room, shifting position uncomfortably, I say “Justin’s bored”. But Justin isn’t bored; far from it. He’s absorbed in what Debbie’s telling him about pipistrelles – although it’s true his new shoes are making his feet ache. We can make mistakes about our own states of mind (of which more in 3.2), but we don’t seem able to make this kind of mistake. Justin might think himself newly fascinated by pipistrelles when in fact he’s newly fascinated by Debbie, but he doesn’t seem able to think himself bored whilst fascinated, or mistake his aching feet for boredom. Within a certain boundary at least,2 our default assumption is that someone level-headed and sincere who stably and confidently self-ascribes some mental state will do so correctly.3 It is the cases where a person gets it wrong about her own state of mind which need a special explanation, not those where she gets it right. As it is sometimes put, if you want to know about S’s states of mind, S is the best person to ask. These are all aspects of self-ascriptions’ first-person authority. Usually, thinking of someone as an authority on some topic goes hand-in-hand with thinking of them as particularly knowledgeable about it. The person whose word we take on trust on questions of particle physics is the particle physicist: she’s the expert. In court, the testimony of an eyewitness is privileged over certain other kinds of evidence: she should know, she saw it happen. Luke and Peter tell each other everything, so if we want to understand Peter’s odd behaviour, Luke is a good person to ask. When sincere, the particle physicist’s claims about particle physics, the eye-witness’s claims about what she saw, and Luke’s claims about Peter’s behaviour are, compared to claims made on these topics by the rest of us, particularly likely to be true. Why? Because the particle physicist, the eyewitness, and Peter’s friend Luke, know a lot about their respective subject-matters. Many sincere self-ascriptions are recognised as being even more authoritative than the statements of experts, and it is natural to understand this first-person authority as reflecting the fact that we are each in a special epistemic position with respect to our own minds: S is especially likely to accurately self-ascribe her own states of mind, because she is especially knowledgeable about them. This natural thought embodies the epistemic approach to understanding self-ascriptions, and to the relationship to one’s own mental states which these ordinarily manifest. But self-ascriptions have a second distinctive feature, which can seem hard to square with the epistemic approach. For epistemically speaking, self-ascriptions are distinctly unlike other forms of assertion. It is from this perspective that the Expressivist thinks we need a non-epistemic account of first-person authority. Rather generally, it is often reasonable to query a speaker’s epistemic credentials when she asserts something. Sometimes we do this from a sceptical perspective (“Do you know that, or are you just guessing?”; “Have you got any evidence?”; “You reckon? Prove it!”), but sometimes we are just curious about how someone found something out, or what evidence she’s going by (“How do you know?”; “What led you to that conclusion?”). Either way, ordinary assertions about contingent matters of fact are typically open to appropriate epistemic questioning of some kind or another. But these kinds of epistemic query or challenge seem out of place or inappropriate in relation to ordinary self-ascriptions. If I say that Abi is feeling sick, you might ask me how I know this – and I might answer, saying (e.g.) “I slipped some deathcaps in her stroganoff” or “Look at her!” But I can’t sensibly answer in these ways if asked how I know that I am feeling sick. Again, if I say that I am feeling sick you might ask me why this is so (why I am feeling sick), but it is hard to make sense of seeking a substantive answer to the question of why I believe that I am feeling sick. Answers like “Because I feel sick” or “Because I am feeling sick” simply repeat my initial self-ascription; they do not explain it. The same goes in other cases: “How do you know that your feet ache?” is vaguely idiotic unless perhaps heard as a facetious way of downplaying one’s interlocutor’s pain (“You think your feet ache now – come back to me when you’ve tried dancing in these heels.”); asking someone who says she wants to go skydiving why she thinks this, is hard to hear as anything other than a convoluted way of asking after her reason for going skydiving, or of dismissing her desire as beyond one’s own comprehension. Yet read literally, the form of words employed makes this a question about her beliefs, not her desires. “What’s your evidence for thinking that Justin is bored?” is perfectly reasonable; “What’s your evidence for thinking that you believe that Justin is bored?” is hard to interpret full-stop. In all of these cases it is (at least) hard to hear these questions as straightforward requests to be provided with the speaker’s epistemic credentials. Self-ascriptions in this way clearly contrast with expert claims like those considered in 2.1. Asking the particle physicist how she knows that the Higgs Boson has been discovered at CERN, for instance, is perfectly reasonable. The question need not be a sceptical or ‘pointed’ one, one which carries the suggestion that she might not know (Austin 1961, 78; see also Roessler 2013, 3). It might only express an open-minded interest in understanding the structure of the scientific evidence. One common response to the observation that self-ascriptions are in this way epistemically asymmetrical with other forms of assertion, is to deny that in order for a person to have formed their belief on some epistemic basis, or as the result of some epistemic method, this basis or method must be reflectively accessible to her.4 But this appeal to epistemic externalism misses the point. To see why, it is instructive to compare my self-ascription “I want to leave the party” to (the externalist’s friend) the chicken sexer’s claim, “this chick is female”. Say we buy the externalist’s claim that the chicken sexer knows that the chick she is holding is female despite being unable to say how her knowledge is epistemically grounded, or (if this is different) how she came by it. Still, it is perfectly appropriate to ask the chicken-sexer how she can tell, or why she believes, that the chick is female. (Granted, such questions become inappropriate once we discover that she can’t answer them, but this is a separate issue.) The chicken-sexer’s ability to make accurate statements about the sex of chicks whilst having no idea how or on what basis she does so strikes us as mysterious, indeed rather extraordinary.5 The case of an ordinary self-ascriber is not like this. From the common sense perspective, there is nothing mysterious or extraordinary about my ability to truly self-ascribe achy feet, or a desire to leave the party, without being able to cite any epistemic method or basis. On the contrary, the strange thing would be for an interlocutor to expect me to cite any such method or basis. In short, epistemic externalism allows one to hold onto the theoretical possibility of knowing that something is the case without being able to explain on what basis or by what method one knows it. But it does not explain why, in the case of self-knowledge, it is out of place so much as to enquire into the knower’s epistemic basis or method. Plausibly, the ordinary inapplicability of such questions to my self-ascription “I want to leave the party” shows not that the epistemic basis for this self-ascription is inaccessible to me, but that it rests on no epistemic basis at all. The foregoing suggests that the epistemic asymmetry between self-ascriptions and other kinds of assertion is rather intractable. The Expressivist's move is to reject the epistemic approach to understanding first-person authority wholesale: if self-ascriptions are first-person authoritative because they express a (special) form of knowledge, then it should be appropriate to epistemically query the self-ascriber. But it isn’t. So we should reject the epistemic approach. Or so goes the thought. The Expressivist therefore opts for a non-epistemic explanation of self-ascriptions’ first-person authority. Key to this strategy is the idea that self-ascriptions are a special kind of speech-act which does not primarily function to describe or report one’s being in the self-ascribed state, or to express a belief or judgment to the effect that one is in it, but instead to directly express the self-ascribed state itself. Expressivists term this special kind of speech-act an avowal. In avowing “My feet ache”, for instance, I ‘give voice to’, ‘vent’, or ‘speak from’ my pain itself.6 Despite their surface similarities, “My feet ache” (or “I want to leave the party”, or etc.) as it would ordinarily be said, and “Justin’s feet ache” (“Justin wants to leave the party”, or etc.), as it would ordinarily be said, do very different jobs. The main job of the latter is to express my view on Justin’s state of mind. The main job of the former is to express not my view on my state of mind, but my state of mind itself; to express my pain (my desire to leave the party, or etc.). On this view, it is not because they express self-knowledge, but because they are avowals, that self-ascriptions are first-person authoritative. The special relationship a person bears to her own mental states is that of being able to express them, and self-ascription is one way of doing so. This is why S herself is the best person to ask about whether she is in some state M: she can directly express M in an avowal. It is because avowals are direct expressions of one’s mental states that there is not the same scope for inaccuracy in avowing as there is in ascribing mental states to others. The strangeness and uncommonness of being both sincere and inaccurate in self-ascribing a mental state can be understood on the model of other kinds of non-epistemic expressive mistake, such as saying “Ow!” on feeling a mild and pleasant sensation, or “Hurrah!” when one means “Oh no!” Expressivists explain self-ascriptions’ epistemic asymmetry in a related way. Rather generally, a person doesn’t need any epistemic basis for taking herself to be in some state in order to express it – to express her pain by groaning, say. When it is an avowal of pain, my self-ascription “My feel ache” needs no more evidential support than does a pained grimace. Nor should we expect my capacity to self-ascribe my pain to depend on any method for discerning it, or process of coming to know about it. On this view, it is because there are simply no answers to epistemic queries about self-ascriptions that the queries themselves turn out to be inappropriate. As we have seen, the Expressivist denies that first-person authority is to be explained in terms of a self-ascriber’s self-knowledge. But this is consistent with claiming that a self-ascriber will typically have self-knowledge of the state she self-ascribes, and even that this is expressed in her self-ascription. In 3.1 I describe three forms of Expressivism, distinguished by their attitudes to these additional claims. In 3.2 and 3.3 I argue that none of these positions is strongly motivated. Understanding what goes wrong for these different forms of Expressivism will motivate the approach to understanding self-ascription which I develop in sections 4 and 5, 4 and 5. I call the three forms of Expressivism I want to distinguish ‘Eliminativist’, ‘Deflationist’, and ‘Parallelist’ Expressivism. What I call Eliminativist Expressivism is often attributed to the later Wittgenstein, and is held by some contemporary Wittgensteinians (Kenny 1973, ch. 10; Hacker 1986, ch. X). As we have already seen, it is central to any Expressivist view that we can speak truly about our own states of mind.7 What the Eliminativist denies is that this capacity is in any way an epistemic one. The Eliminativist does not claim that we are ignorant about our own states of mind, but more fundamentally that epistemic concepts simply don’t apply here. Our capacity to speak truly about our own minds is expressive rather than epistemic, so that describing a person either as knowing or as being ignorant about her own belief, pain, etc., would be equally, strictly speaking, nonsense.8 Deflationist Expressivists take their mark from a weaker reading of Wittgenstein (Finkelstein 2003 §6.4; see also Kemmerling, 2012). Unlike the Eliminativist, the Deflationist allows that one can “meaningfully [and] truly” (Finkelstein 2003, 151) describe someone who avows some mental state as expressing knowledge that she is in it. But she also denies that this ‘knowledge’ is knowledge, in an ordinary or substantive sense. For the Deflationist, prefacing a statement like “I want to leave the party” or “My feet are aching” with “I know” makes no substantive difference to its meaning or truth-conditions. Only those who I call Parallelist Expressivists wholeheartedly accept that avowals like “My feet ache” or “I want to leave the party” express a genuine form of knowledge (see Bar-On e.g. (2004)). The Parallelist, like all Expressivists, rejects the epistemic approach to understanding first-person authority, explaining the latter in terms of expression and avowal. But she also holds that we genuinely know facts about our own states of mind, and that this knowledge is expressed in our self-ascriptions – as it were, in parallel to the expression of the self-ascribed state. For the Parallelist, self-ascriptions are first-person authoritative, and genuinely knowledgeable, but not first-person authoritative because they are genuinely knowledgeable. I start with a specific worry about Deflationist Expressivism. As noted above, the Deflationist holds (a) that it can be both meaningful and true to say that someone knows that (e.g.) her feet are aching, but (b) that doing so adds nothing substantive to saying that her feet are aching. Holding (a) distinguishes the Deflationist from the Eliminativist, whereas holding (b) distinguishes her from the Parallelist. But intuitively (a) and (b) aren’t a happy combination. If attributions of self-knowledge are meaningful, how can they add nothing to the meaning of simple attributions of the relevant (i.e. putatively self-known) states? to into either the epistemically weaker or the epistemically But I on this. Because even if I do not think that it is and for the same reason that I think that is not In short, I am not by the considerations for that our to the states we self-ascribe are genuine or substantive If we are the word know” as it is how are we to then other people very often know when I am in but all the same not with the with which I know it It be of me at all perhaps as a that I know I am in What is it to perhaps that I am in people be to of my only from my behaviour, for I be to of them. I have them. The it makes sense to say about other people that they whether I am in but not to say it about are to the kind of it only makes sense to apply the ordinary knowledge in where it would also make sense to apply certain other related epistemic concepts like and But it doesn’t make sense to think of someone as having or as that they are in a certain mental state. So either it doesn’t make sense at all to say that a person knows that she’s in some mental state the Eliminativist or it can make sense to say that someone knows that she’s in some mental state, but only if we in a or sense the Deflationist for the of that it ordinarily make sense to say that someone has that they are in some mental state, or that they might whether they are in some Still, this doesn’t us very For there a of other concepts which are related to knowing which can sensibly and be to a relationship to facts about her own mind. This true even in relation to facts about one’s own and sensations, states often as especially to being of genuine first-person knowledge. So I on it’s to make mistakes or it right about whether one is in some mental It is familiar to mistake a in one’s for a pain when in fact one has it up the cases also that, one can or to that one is in some state – in this I come to that I am having a of in my rather than the of up or in this is an of of a sensation, rather than of developing (e.g.) a or feeling as one it’s to be whether or that in some mental state, for at its being whether a certain is a pain or an and then after a that it’s (e.g.) a Again, I don’t deny that there are cases in which what as an a pain and often seem to be like this – from to pain as the or cases where a is The is that not all cases are like this. particularly or person interpret as something which they later to be This kind of thing strikes me as common in need of a special it to in some state so happy to see I all about how much my feet and to being in it so and I how much my feet and to we of facts that we know – and no of facts about our own states of mind. we from facts that we know, and no from facts about our own states of and so on. In even if we that some epistemic concepts related to that of knowledge do not apply to our to facts about our own mental it true that other such concepts do so I don’t want to deny that there are between the ways in which epistemic concepts such as mistake, being etc., might our to facts about our own mental by contrast with the ways in which they our to facts about the mental states of or But this would only us the that self-knowledge is in certain different from knowledge of facts about other minds and about I think that this is and a of the account I in 4 is to accommodate and explain it. But the of either Eliminativist or Deflationist Expressivism. The Parallelist Expressivist that self-ascriptions ordinarily manifest a substantive form of knowledge. But also a of is a for the Parallelist, which when we ask whether or not self-knowledge is like other forms of knowledge, to epistemic support by or method. For I the in relation to evidence. my knowledge that I want to leave the this rest on evidence for thinking that I want to If she says that it the Parallelist will be without an explanation of its epistemic asymmetry with other kinds of assertion. The central Expressivist is that a non-epistemic account of self-ascriptions’ first-person authority can explain their epistemic But if the Parallelist of my self-ascription “I want to leave the party” as expressing self-knowledge that I want to leave the party, in to expressing my desire to then the question of why it should be inappropriate for you to ask me to cite my evidence simply The explanation of my epistemic asymmetry with other assertions will have The Parallelist is therefore likely to that the self-knowledge expressed in a self-ascription is The fact that my self-ascription knowledge that I want to leave the its epistemic asymmetry with other forms of assertion if the knowledge it is itself epistemically asymmetrical with that expressed in other forms of assertion, in of unlike them, But that now the Parallelist has no reason to offer a non-epistemic account of first-person authority. The for doing so – – to accommodate self-ascriptions’ epistemic asymmetry with other forms of assertion. But if the self-knowledge expressed in a self-ascription is already unlike the knowledge expressed in other in being then an explanation of self-ascriptions’ epistemic asymmetry for and there is no for a non-epistemic account of first-person authority to the of a non-epistemic account of first-person authority terms of the phenomena of expression and is the of Expressivism, Parallelist Expressivism is As I explained in it can seem for an epistemic approach to understanding first-person authority to accommodate self-ascriptions’ epistemic asymmetry with other forms of assertion. It is from this perspective that the kind of non-epistemic approach by the Expressivist to But on no form of Expressivism able adequately to capture self-ascriptions’ epistemic The for that self-ascriptions express a substantive form of knowledge is not so I think we should reject both Eliminativist and Deflationist Expressivism. The Parallelist allows that self-ascriptions express genuine self-knowledge, but in so doing her Expressivism. On one of the in 3.3 she is unable to explain self-ascriptions’ epistemic on the other she the need for a non-epistemic account of their first-person authority. Yet the central claim of Expressivism is that we need a non-epistemic account of self-ascriptions’ first-person authority in order to explain their epistemic I want to develop an account of self-knowledge and self-ascription which is the of that on the second of the This to the Expressivist because it is a version of the epistemic approach. But if we are to that self-knowledge is genuinely epistemically I think that the epistemic approach out to be I develop these in the rest of the to the epistemic approach, self-ascriptions are first-person authoritative because they are particularly likely to express knowledge of one’s own mind. But if the epistemic approach is to it needs to do to the epistemic asymmetry between self-ascriptions and other forms of assertion. as as we are not to make certain about the of knowledge in and self-knowledge in I think that it is not only natural but also to view first-person authority in epistemic The main assumption that we need to is that knowing that is rather a of having a belief that This very which on some very what true belief must be in to as knowledge epistemic process or whether is epistemically or and whether or not knowledge is to true So the of knowledge as true belief is very But it is far from this is we can start to how it might be and how it might make for a understanding of the distinctive of both self-knowledge and self-ascription. This is my for the rest of the explanation of why self-knowledge is not epistemically so of self-ascriptions’ epistemic rests on the idea that simply being in a mental state is in for knowing that one is in it. means inter without the of additional reasons or evidence for thinking that one is in will say more about this idea in person has self-knowledge as we have been understanding it – epistemically ungrounded knowledge that one is in some state of mind – when she knows about her own state just in of being in This both self-ascriptions’ first-person authority and their epistemic as The idea first-person authority in the kind of epistemic terms which the Expressivist us – I have – to It allows us to say that sincere self-ascriptions are strongly or by default to be and that the best person

Assertion · Epistemology · Expression (computer science · Expressivism · Natural (archaeology · Self-knowledge · State (computer science · Computer Science · Epistemology, Ethics, and Metaphysics · Feminist Epistemology and Gender Studies · Philosophy · Philosophy and Theoretical Science · Psychology

• Self-Knowing Animals

  Open Access•Evan Jones•Philosophical Explorations•2026

  Add to list

• Expressing the Absence of Emotion

  Open Access•Nadja-Mira Yolcu•Review of Philosophy and Psychology•2026

  Add to list

• Self‐Knowledge and the Capacity to Judge

  Open Access•Matthew Parrott•European Journal of Philosophy•2026

  Add to list

• The Standing of Self-Identity

  Open Access•Kenneth Silver•Hypatia•2026

  Add to list

• No-‘How’ Privileged Self-Knowledge

  Open Access•Dorit Bar‐on•Erkenntnis•2024

  Add to list

• Action, Knowledge, and Will

  John Hyman•Action, Knowledge, and Will•2015

  Add to list

• Speaking My Mind

  Dorit Bar-On, Dorit Bar‐on•Speaking my mind•2004

  Add to list

• A Virtue Epistemology

  Ernest Sosa•A virtue epistemology•2007

  Add to list