Most philosophers, and in my experience, many non-philosophers as well, have the intuition that there is something distinctive about the knowledge we have of our own mental states. If you were to tell me that you think that there is beer in the fridge, I might ask you why you think that, but it would border on pedantic for me to ask you how you know that that is what you think. We tend to grant a kind of authority to first-person reports on one’s own mental states. Indeed, we tend to assume that our access to our own mental states is unique in at least two ways. First, the access we have to our own mental states is said to be “privileged” in the sense that the second-order beliefs that we have about our own mental states are more likely to amount to knowledge than our beliefs about, say, everyday objects or other minds. This access is said to be “peculiar” in that the method by which one comes to know one’s own mental states is often a method that is not available to others in the same way it is available to oneself. However, explaining these purported features of self-knowledge proves to be a difficult task. Though numerous accounts of self-knowledge have been put forward, over the last two decades, so-called “transparency accounts” of self-knowledge have emerged as some of the more promising for explaining the privileged and peculiar access we have to our own mental states. These accounts maintain that one can come to know one’s mental states not by attending to the mental state itself, but by attending to the world ‘outside’ of one’s mind.
Transparency accounts take their cue from a handful of admittedly cryptic passages in Gareth Evans’s The Varieties of Reference. Evans writes that “in making a self-ascription of belief, one’s eyes are, so to speak, or occasionally literally, directed outward-upon the world.” He then goes on to observe, “If someone asks me ‘Do you think there is going to be a third world war?’, I must attend, in answering him, to precisely the same outward phenomena as I would attend to if I were answering the question ‘Will there be a third world war?’” (p. 225). If asked whether or not I believe that there will be a third world war, it seems I can self-ascribe the relevant belief (e.g., “I believe that there will not be a third World War”) merely by considering some specific facts about the world (e.g., facts regarding the political state of affairs in eastern Europe).
Evans’s suggestion is exciting for a number of reasons. First, Evans’ account is well-positioned to explain privileged access. The second-order beliefs that have been formed in the way he describes are very likely to amount to knowledge. If, when asked whether I believe there will be a third world war, I consider what I take to be the relevant facts and respond, “I believe there will be a third world war,” this self-ascription seems very likely to track my first-order belief. Second, this procedure appears to be peculiar, i.e., distinctively first-personal. Though your assessment of the political situation in Eastern Europe might reveal whether you believe that there will be a third world war, it tells you nothing, or at best very little, about what I believe on this subject. Finally, the view is economical. That is to say, it does not refer to any proprietary introspective faculty and thus avoids many of the problems of standard inner-sense accounts.
At this point, you would be forgiven for feeling a bit uneasy about Evans’s claim. The suggestion that this procedure produces self-knowledge at all is something of a puzzle since the kind of evidence that would usually justify an answer to a question like “will there be a third world war” is quite distinct from the kind of evidence that would justify a claim about what someone believes. It does not seem obvious that looking at non-mental facts related to the political turmoil in Eastern Europe signifies anything about one’s beliefs. Evans himself does very little to explain why the self-ascriptions that follow from this procedure constitute knowledge, or to explain how this can be a successful model for self-knowledge more generally. Indeed, it is unclear that Evans thought this method was viable for mental states other than belief.
Nonetheless, numerous attempts have been made to develop Evans’s insight into a generalized account of self-knowledge, and these transparency accounts take a variety of forms: Akeel Bilgrami, Matthew Boyle, Alex Byrne, Annalisa Coliva, Jordi Fernández, Richard Moran, and Antonia Peacocke have all offered distinctive accounts of transparency. Though this is not the forum for articulating each of these accounts, two have been particularly influential and are thus worth exploring in more detail. Let’s start with the “epistemic accounts” of the sort proposed by Alex Byrne and Jordi Fernández. Epistemic accounts take Evans’s claims to be descriptive of a genuinely epistemic phenomenon. Though they differ in their details, epistemic accounts attempt to explain self-knowledge by reference to an empirical justification for the self-attribution of the first-order state. Whatever is unique about transparent self-knowledge, these views claim, can be explained in standard epistemic terms. For example, Byrne claims that it is by following a schema such as Bel—“If p, believe that you believe that p”—that one acquires knowledge of one’s beliefs. The key move consists in claiming that if one recognizes that p is the case, then one will ipso facto form the belief that p. Bel is thus “self-verifying” since if one follows Bel, one will form the belief that one believes that p. Furthermore, Bel is “strongly self-verifying” since even if p is not the case, but one mistakenly thinks that p is the case, one will still form the belief that p.
Despite the appeal of Byrne’s account and epistemic accounts more generally, they have their share of flaws. Looking again at Byrne’s inferential strategy, one might note that if the basis for the inference were one’s knowledge of having accepted that p is the case, that would presuppose that one already knows one’s own mind, and the account would be circular. Alternatively, if the starting point of the inference is p itself and not one’s representing oneself as believing that p, then the inference would be, in Mathew Boyle’s words, “mad.” For, if one asks oneself why one believes that one believes that p, the answer “p” is woefully inadequate; even if p is true, this by itself need not support the claim that one believes p. Thus, Byrne’s account faces a dilemma: it is either circular or such as to impute an “mad” inference to the subject. These considerations have led many to think that transparent self-knowledge cannot be grounded in inference.
In opposition to these epistemic accounts, Richard Moran argues that transparent self-knowledge is explicable in terms of our nature as rational agents rather than any empirical justification. Central to Moran’s thesis is a distinction between “deliberative questions” and “theoretical questions.” According to Moran, a theoretical question is a question that is answered through the discovery of some fact or evidence of which one was previously ignorant, while a deliberative question is answered not by discovery but by decision. So, for example, while casually surveying the menu at your favorite restaurant, you might ask yourself what you will have for dinner. The answer to this question will not ultimately be decided based on some discovered fact—after all, you know the menu well enough. Instead, the answer will be determined by what you decide to order. Compare this to asking yourself what you will have for dinner as a guest at a dinner party. This question will only be answered by a discovery of sorts; namely, it will be answered by finding out what the host is cooking up in the kitchen.
Moran argues that, in the cases of transparent self-knowledge, the theoretical question as to what one believes is overtaken by the deliberative question as to what one ought to believe. But, Moran notes, determining the answer to this latter question is simply a matter of determining what is true. As such, when asked whether I believe that p, I should be able to report on my belief simply by considering some specific facts about the world, namely, whether p is true. That my report, when formed in this way, will accurately track what I believe is constitutive of the practice of rational deliberation. For, Moran argues, if one takes it as an open question whether the activity of deliberation determines what one actually believes, then the very activity and aim of deliberation is undercut.
Though most agree that Moran is on to something here, his account notoriously has trouble explaining how such self-ascriptions can amount to knowledge. It has also been argued that Moran over-intellectualizes the nature of self-knowledge; for, Moran’s account seems to require that the self-knowing agent is in possession of a specific set of psychological concepts (e.g., the concept of belief) as well as some concepts essential for the practice of deliberation. It would follow from this that individuals lacking such concepts, for example, young children, would necessarily lack a certain kind of access to their own mental life. In response to these concerns, Moran has suggested that self-ascriptions made via transparency constitute knowledge insofar as they rest on an entitlement attained through the possession of certain rational capacities. This entitlement, Moran claims, is available irrespective of one’s conceptual proficiencies. Unfortunately, as intriguing as this suggestion is, Moran himself does not develop it in any detail.
It is safe to say that these two accounts have done more to shape the discourse on transparency than any of the others. Not so much because people find either account to be overwhelmingly convincing, but more because their limitations help to define the contours of the problem. I am inclined to think that our ability to self-ascribe beliefs via transparency does derive from our proficiency with a truth-normed concept of belief. Moran, in my opinion, was too quick to distance himself from this idea in his defense against the charges of intellectualism. I do not think that it is an accident that transparency accounts are most plausible when developed as an explanation for how we come to know our own beliefs, rather than, say, our intentions or perceptions. Nor do I think it an accident that transparency accounts can only explain how we come to know our beliefs qua commitments. The transparency method fails outright as a means for coming to know our beliefs qua dispositions. Indeed, it seems to me that a self-ascription of the belief that p can be substituted for an assessment of the way one takes the world to be, precisely because our concept of belief is such that a belief in p is appropriate only if p is true. Transparent self-knowledge is best understood as the result of a capacity for modulating what one takes to be true of the world into the register of belief, and this capacity is likely part and parcel of our gaining a proficiency with a truth-normed concept of belief. Elsewhere, I have tried to argue that the self-ascription in these cases can be justified along broadly coherentist lines insofar as the second-order belief will conform to a number of the agent’s other beliefs regarding (i) the state of affairs relevant to p (ii) the nature of evidence sufficient for belief generally, and (iii) their understanding of themselves as a rational agent. But this story will likely seem unsatisfactory for reasons all its own.
The debates between various transparency theorists can seem esoteric, but they are, I believe, quite valuable. In tracking these disputes, several serious questions arise not only regarding the nature of self-knowledge but also regarding specific issues in epistemology and philosophy of mind. Rather than providing a full-fledged defense of my own view, I’ll conclude by simply highlighting a few of these questions. First, this debate has brought about a somewhat uncomfortable discussion regarding what it means to take some content p to be true. If transparency is seen as a transition from taking p to be true, understood either as an awareness that p is serving as the premise for an inference (cf. Byrne) or an awareness that p is the conclusion of one’s deliberation (cf. Moran), to the self-ascription of the belief that p, how can transparency accounts possibly avoid circularity? For, any account of transparency seems to presuppose that we are at some level aware that we have taken p to be true, and analyzing this in terms of transparency would surely be a fool’s errand. Yet, if one simply posits some brute awareness of having taken p to be true, we might worry that this would undercut the project of transparency. Second, as I have already noted, even if transparency accounts are plausible for belief, it is not obvious that they can be extended to make sense of how we come to know other mental states. This does not bode well for a unified theory of self-knowledge and might suggest that what we call “self-knowledge” is not a singular phenomenon. Indeed, the more time I spend trying to understand transparency, the more confident I feel in the correctness of a pluralist approach to the study of self-knowledge. Finally, many philosophers see transparency accounts as the most plausible means for explaining privileged and peculiar access. If these accounts should then be found wanting, we would be at a loss to explain our intuitions that there is something distinctive, from an epistemological standpoint, about self-knowledge.
One of the curious appeals of studying transparency accounts is that there seems to be just as much to be learned from their failure as there is from their success; so, whether you endorse or inveigh them, I hope you’ll agree that they deserve your attention.
Edward Mark
Edward Mark is an instructor of philosophy at Loyola Marymount University. He works primarily in the areas of epistemology and philosophy of mind with a particular focus on self-knowledge. He received his PhD from the University of California, Irvine in 2024.
