First-Personal Self-Knowledge

The extent and interest of third-personal self-knowledge notwithstanding, first-personal self-knowledge too deserves attention. In The Varieties of Self-Knowledge three chapters are devoted to a critique of contemporary accounts of it. In particular, I consider Armstrong’s reliabilist model, Peacocke’s and Burge’s different kinds of rationalism, Evans’s transparency method and its two most developed heirs—Moran’s and Fernández’s accounts—as well as Bar-On’s expressivism. There is no room here even to summarize those parts of the book. What emerges, though, is the difficulty of considering first-personal self-knowledge as the result of any kind of however minimal epistemic procedure. For any epistemic procedure may not be executed, or else, be executed and go wrong, without thereby compromising a subject’s concepts’ possession or rationality. By contrast, if a subject, who had just hit their elbow badly against the wall, winced and moaned like the rest of us, and yet did not assent (at least) to “I am in pain”, that would cast doubt on the fact that they possess the relevant concepts. Or else, if, in the same kind of scenario, there were reasons to still grant them those concepts, and yet they said “I am not in pain” that would cast doubt on their being rational. For, ex hypothesi, they do have a painful experience, they exhibit its characteristic bodily behavior, and they do have the relevant concepts. Their self-ascription, then, could only be interpreted as a sign of perhaps momentary madness.

For these reasons, I side with so-called constitutive accounts of first-personal self-knowledge. Accordingly, it is constitutive of being a subject endowed with the relevant concepts, cognitively well-functioning and rational, that if one has (the relevant kind of) mental state M, one will judge that one does and, if they does so judge, one will have that mental state.

Constitutive theorists are united in holding this thesis and in their anti-epistemicism regarding first-personal self-knowledge. A frequent critique is that they seem to fail to explain precisely what they set out to explain, namely how we can have first-personal knowledge of our own mental states. This is an unfair objection, though. There is no real pre-theoretical reason to think that whenever we employ the term “knowledge” we are committed to there being, roughly, true and justified belief. Knowledge-how, by most theorists’ lights, would not qualify as knowledge if that were the case. Similarly for the kind of knowledge expressed by sentences such as “John knows Mary”, in the sense of being acquainted with her, and of being able to single her out (which, if John were a puppy and Mary its mother, could hardly be interpreted as John’s having true and justified beliefs about Mary). Furthermore, it may just be an ingrained philosophical habit, instilled in us since Descartes’ Meditations at least, to think that psychological avowals have to be underwritten by some kind of epistemic procedure, which would result in the relevant judgements’ being transparent and authoritative. Finally, according to a different philosophical tradition, tracing back to the later Wittgenstein, there are some employments of “know”, in various contexts of use, which do not express the obtaining of any specific epistemic relation between a subject and a proposition (or a fact), but should rather be taken as “grammatical”. That is, as expressing the idea that it is constitutive of, roughly, being a subject endowed with the relevant concepts, under certain already specified conditions, that if one has a given mental state M one will judge that one does, that and if one so judges, one will have that mental state.

What remains to be clarified is, first, the range of mental states for which the constitutive thesis holds; secondly, the kind of a priori argument one can give in favor of that thesis for the relevant class of mental states; and, finally, its metaphysical consequences.

Here I will very briefly summarize the gist of the last part of the book. In my view, the constitutive thesis holds only in a limited amount of cases. It holds for self-ascriptions of occurrent sensations like pain, occurrent basic emotions like fear, occurrent perceptions like seeing an apple in front of one, self-verifying psychological self-ascriptions like “I am thereby thinking that it is cold”, and for self-ascriptions of commissive propositional attitudes, like “I intend to spend more quality time with my kids” or, after considering reasons for and against, “I believe politicians should make the right decisions, not just the most popular ones” or “I want my kids to have demanding yet understanding teachers”.

The way the constitutive thesis can be redeemed varies depending on the kind of mental state at issue. So, for instance, while it is constitutive of, among other things (cf. supra), being a subject capable of enjoying bodily sensations such as pain (or basic emotions like fear), that if one is in pain one will so judge (and vice versa), the same does not hold for perceptions. For perceptions can occur and yet we may not be conscious of them (like in the case of blind-sight). Hence, I claim that it is constitutive of, among other things, being a subject capable of enjoying perceptions, which would enter the explanation of actions one could rationally be held responsible for, that if one sees that P, one will so judge (and vice versa).

The case for the recognition of the existence of commissive propositional attitudes cannot be summarized here, but there is a growing consensus that they exist and that they are interestingly different from their purely dispositional counterparts. Common to all accounts that accommodate their existence is the claim that they are based on the (at least potential) assessment of reasons in favor of their content. Furthermore, there is a growing consensus that they are such that subjects could be held rationally responsible for having them, or for not bringing their behavioral dispositions in line with them, should that occur. So I hold that it is constitutive of (among other things) being a subject capable of enjoying commissive propositional attitudes that, if one has them, one will so judge (and vice versa).

The recognition of this class of propositional attitudes allows us to dispense with self-deception as a counter-example to authority. On this account, self-deception arises when there is a conflict between one’s commissive propositional attitudes and one’s purely dispositional ones, such that one judges that one has that commissive propositional attitude while behaving in ways that run contrary to it. Accordingly, subjects’ self-ascriptions of their commissive attitude are correct and authoritative, while their purely dispositional mental states cause them to act in ways that run contrary to their commissive attitudes.

This has a bearing also on the understanding of Moore’s paradox. For, once one is hospitable to the distinction between commissive and purely dispositional mental states, it may so happen that, through third-personal methods, one realizes that one believes (dispositionally) that P, while at the same time sincerely judging that not-P is the case. In such a case, Moorean judgements (or sentences) would no longer be paradoxical, however rare the cognitive situation they may be used to depict might be. It is part of my view that if one really wants to save Moore’s paradox, then one has to interpret the doxastic conjunct in “I believe that P, but it isn’t the case that P”, as an expression of one’s belief as a commitment that P. The crucial feature of beliefs as commitments is that they are intrinsically normative mental states. That is to say, one could not have them while also knowingly and willingly assenting to their negation. By contrast, with beliefs as dispositions, one could have them while also knowingly and willingly assenting to their negations.

In future work I hope to be able to explore the deep consequences of this account of belief for belief-revision, the possibility of intra-personal disagreement, the nature of normativity and the grounds of the principle of non-contradiction.

Finally, let me close by stating what I take the metaphysical implications of the constitutive thesis to be. Small surprise that, even in this context, I have a pluralistic view. For, it seems to me that only in the case of commissive propositional attitudes (and of self-verifying psychological self-ascription) one’s self-ascription can actually bring the first-order mental state about. In the case of basic sensations, basic emotions and perceptions, in contrast, their self-ascription serves a different function. Namely, it merely individuates them for the mental states they are (with the content they have).


  1. V. P. P.

    Excellent piece. I wonder if you could clarify a bit more exactly what self-knowledge amounts to on the constitutivist picture. I get that one doesn’t have to have a JTB. But what does one have to have? What is the nature of the mental state one is in when one has self-knowledge? Is my knowledge that I believe that p simply a component of my commissive belief that p? Or is it produced by some mechanism that cannot fail that one must possesses if one has the commissive belief that p, securing that belief that p is always accompanied by knowledge that one believes that p?

  2. Annalisa Coliva

    That’s a very good question, thanks!
    I think we should distinguish various possibilities.
    First there is forming the commissive belief by just going over reasons for P and judging that P (because one takes – rightly or wrongly – those reasons to be sufficient to establish P). In that case, one has the commissive belief that P.
    Commissive propositional attitudes are therefore possible independently of their self-ascriptions.
    The self-ascription, however, is possible and it should be explained in a way that does not presuppose either an observational or an inferential account of how it comes about, at least in paradigmatic cases.
    In the book, I propose to endorse the expressivist story to some extent. So the idea is that we are actually taught to replace the immediate manifestaion of that belief (i.e. the assertion of P) with “I believe that P”. Once we have acquired that ability, we can indifferently express our commissive belief either through asserting that P, or through asserting “I believe that P”.
    Finally, we can sometimes form the commissive belief that P, after going through reasons for P and having taken them to be sufficient to establish that P, through its very self-ascription.

  3. Annalisa Coliva

    Let me clarify a bit more. There is just one mental state – the commissive one. Depending on one’s conceptual resources, that state can be expressed differently – either through the assertion of P, or of “I believe that P”. But it is one and the same state. When dressed in psychological terms, it can actually play a more encompassing functional role.

    • V. P. P.

      Thank you, this was very helpful. But I still don’t quite understand why self-knowledge isn’t a separate kind of mental state, different from the commissive belief. The consitutivist claims that it’s constitutive of being a rational subject etc. “that if one has (the relevant kind of) mental state M, one will judge that one does”. So when one judges that one does, isn’t the judgment a separate mental act? One judges and commits oneself to it being the case that one has M. We often token such beliefs, and they’re just like first-order beliefs, just that they’re about our own mental lives. They can play the role of premises in inferences (e.g. ‘I believe that Hillary is president, so I believe that a woman is president’), guide action (without one tokening the commissive belief that p) etc. And of course, when I have the belief that I believe that p, the consitutivist is committed to it being the case that I’ll also judge that I have the belief that I believe that p, and so on. Are commissive beliefs that I believe that p – which we no doubt sometimes token – always formed through inferential or other methods on the constitutivist picture? I agree with your claim that there is no “pre-theoretical reason to think that whenever we employ the term “knowledge” we are committed to there being, roughly, true and justified belief.” But surely people often do form justified+ true beliefs about their own mental states, and often when we claim that a person has self-knowledge, we mean to claim that the person has just such a belief (alternatively knowledge that she believes that p, where knowledge is understood as a sui generis mental state).

  4. Annalisa Coliva

    Thanks a lot for pressing this point. Now, I do not deny that sometimes we report on our commissive mental states. So suppose I notice that my son has a weird behavior and considering various factors I come to the conclusion that it would be good to spend some time just the two of us together. So I deliberate “I intend to take him for lunch just the two of us”. This is a psychological self-ascription and one by means of which I actually form the commissive mental state (the intention of taking him out for lunch). The of course I can rememebr it and reason “If I intend to take him to lunch, I will have to find a quiet restaurant where we can talk”, etc. But in this kind of reasoning I am not deliberating and forming the commissive mental state. Rather, I am reporting on it, based, presumably, on mnestic evidence. The constitutive thesis is supposed to take care only of the deliberative step, not of the reporting one.

  5. Annalisa Coliva

    No a physical structure does not have knowledge even less is knowledge.

    A gene carries information, like the rings of a tree.

    Philosophy helps us learn how to make finer distinctions, which would be useful to science too. Proably not to its development, but to how people portray its results to the general public.

Comments are closed.