Evans, transparency, and Cartesianism

Correspondence David Zapero, University of Bonn, Institut für Philosophie, Am Hof 1, Bonn 53113, Germany. Email: dzaperom@gmail.com Abstract In The Varieties of Reference, Evans makes two parallel claims about thought and perception. He argues that both our capacity to self-ascribe thought and our capacity to self-ascribe perception are fallible. The essay focuses on his claim about perception and examines its relation to Evans's project of rejecting a Cartesian conception of the mind. In his theory of perception, I argue, Evans embraces a conception of first-person authority that he seeks to reject in his account of thought. He is thus not able to extend to perception the critique of Cartesianism that he develops in the context of his analysis of thought. My intent is not exclusively critical. By working through the difficulties that Evans's argument raises, I seek to shed light on whether and, if so, how criticisms of a Cartesian conception of thought may be extended to the philosophy of perception.

about the fallibility of our capacity to self-ascribe thought, is closely related to another claim about our capacity to self-ascribe belief. Generally, Evans argues, we do not relate to our own beliefs by way of an "inward glance"; we instead arrive at a self-ascription by "looking outward" and directing our attention to what our beliefs are about. 2 Evans defends these claims as part of his endeavor to reject a Cartesian conception of the mind, and he makes parallel claims about perceptual experience. That is, Evans also argues that our capacity to self-ascribe perceptual experience is fallible and that we arrive at self-ascriptions of experience by directing our attention to what our experience is about. Yet, while Evans's claims about how we relate to our own thoughts and beliefs have been widely debated, his conception of our relation to our own perceptual experience has received significantly less attention. In the present essay, I examine that conception and its relation to the project of rejecting a Cartesian conception of the mind.
There are two points I want to make about Evans's position. Firstly, his account of our capacity to self-ascribe perceptual experience makes such self-ascriptions immune to a certain kind of mistake. Evans can allow for erroneous self-ascriptions, but he ends up crediting subjects with an epistemic authority that is not unrelated to the authority characteristic of an infallible capacity. Moreover, by granting subjects such an epistemic authority, Evans embraces in the theory of perception a conception of first-person authority that he seeks to reject in his account of thought. Contrary to what Evans suggests, he is not able to extend to perception the critique of Cartesianism that he develops in the context of his analysis of thought.
My intent is not exclusively critical. By working through the difficulties that Evans's argument raises, I seek to examine whether and, if so, how criticisms of a Cartesian conception of thought may be extended to the philosophy of perception. I seek to bring out a certain constraint that must be satisfied by any account of perceptual experience wishing to extend those criticisms. It is a constraint, which, notably, many intentionalist theories of perception, including Evans's own version of such a theory, do not satisfy.

| THE ELUSIVENESS OF SENSORY CONSCIOUSNESS
I will begin by taking a bit of a detour and examining Moore's influential claim according to which sensory experience is transparent or diaphanous. Contrasting his claim with Evans's claim will allow me, later on in the essay, to bring into view more sharply the feature of Evans's conception of experience, which leads him to credit subjects with a very particular epistemic authority.
In the context of his attempts to determine the nature of sensations, Moore famously makes the remark that "in general, that which makes the sensation of blue a mental fact seems to escape us: it seems, if I may use a metaphor, to be transparent-we look through it and see nothing but the blue." 3 Not much later in the same essay, he also states: "the moment we try to fix our attention upon consciousness and to see what, distinctly, it is, it seems to vanish: it seems as if we had before us a mere emptiness. When we try to introspect the sensation of blue, all we can see is the blue: the other element is as if it were diaphanous." 4 These and other remarks on the transparent or diaphanous character of consciousness from the second half of "The Refutation of Idealism" have come to be associated with intentionalist accounts of perceptual experience. For various contemporary authors defending such accounts of perceptual experience have felt that Moore's considerations allow to ground the central tenet of intentionalism, namely, that the nature of sensory experience can be explained in terms of the representational work that such experience achieves. 5 In the aforementioned passages, Moore is understood to be saying that when we try to introspect a perceptual-or at any rate a visual-experience occurring in normal circumstances, we are not afforded awareness of the experience or of its qualities but, instead, of the qualities or objects in our surroundings. When trying to focus on our experience-Moore is taken to be claiming-, we do not bring into view something different from what that experience is an experience of; we are instead led right back to the external world which that experience represents. 6 It is however, far from clear that Moore's remarks can be understood in this way. 7 Indeed, his remarks are, in one sense, and as he himself insists, quite unambitious. For they are not meant to establish the epistemological thesis that sensory experience (normally) puts us in touch with our environment or the external world. Rather, by appealing to introspection, Moore seeks to make the case that the nature of an episode of sensory experience must be specified with reference to an object with which the subject of such an episode is presented. The remarks about transparency involve no claim about the status of such an object, nor a view as to whether that object is a part of our shared environment. Moore does at first defend the position that such objects-what from 1909 onward he calls "sensedata"-are mind-independent, but nothing in the passages quoted above guarantees that they will even be that. In those passages, Moore works with a picture of sensory experience according to which the nature of such experience is determined by an object to which we entertain the generic relation of "consciousness" or "acquaintance." 8 In fact, in his essay Moore defines "object" simply as that with respect to which one sensation differs from another, so it is somewhat misleading to present his claim as I have just done. 9 If we remain loyal to his terminology, my framing of his claim would amount to a tautology. It is thus preferable to start off by putting his claim in the following terms: sensing involves the presentation of an object or a quality, which must exist for that episode of sensing to be the one it is. 10 Of course, such a reframing of the claim also only gets us so far, since we are now confronted with a quite indeterminate notion of existence. One of the major difficulties that Moore and other partisans of his approach faced was to explain just what existence or instantiation amounts to in the context of such a claim about sensory experience.
One option is to elucidate these notions by drawing on the idea of a publicly observable environment. "Instantiate" or "exist" would then simply refer to the fact that something is part of our shared environment and available to be consciously apprehended by a suitably equipped observer. On this understanding of "instantiation," the transparency claim would amount to the idea that the specificity of a given sensory episode-what distinguishes it from any other such episode-must be specified with reference to an object or state of affairs in our environment with which the episode puts us in touch. Instantiation and existence would thereby be relatively unproblematic notions. But, of course, Moore and the other sense-datum theorists rejected this option and opposed the view-naïve realism-to which it leads. I leave aside here the reasons that led them to this rejection (they were generally grouped together under the heading of "the argument from illusion"). Moore and the sense-datum theorists were, at any rate, confronted to the notorious problem of making sense of the idea that the objects of sensory experience should exist without being a part of a shared environment.
The crucial point for our purposes is that Moore's claim about the transparent or diaphanous character of consciousness does not involve in itself a stance on these issues. Sense-datum theorists were quick to reject naïve realism (the view that the objects of our environment are at least partly constitutive of our sensory experiences), but the transparency claim by itself does not warrant that rejection. The transparency claim only presupposes a certain relational conception of sensory experience according to which every episode of sensory experience must be understood as being, essentially, a presentation of something-of something which is thus constitutive of the nature of that sensory experience. By introducing a relational conception of sensory experience, Moore does not yet decide whether or not the objects with reference to which one specifies the nature of sensory experiences are environmental ones. Indeed, Moore introduces the term "sense datum" precisely to remain neutral on this matter: sense data are simply whatever is given or presented in sensory experience, that is, that with reference to which the nature of such an episode is specified.
In the context of this relational conception of sensing, the transparency claim is the claim that the nature of a sensory experience is entirely determined by the relevant object. Which is why it can also be framed as a denial-a denial that it is possible to distinguish between different kinds of relations. This is indeed how Price presents the claim at the beginning of his Perception: Are there several different sorts of acquaintance, e.g. sensing, self-consciousness, and contemplation of mental images? I cannot see that there are. The difference seems to be wholly on the side of the data. If so, a fortiori there are not different kinds of sensing. Visual sensing will simply be the acquaintance with colour-patches, auditory sensing the acquaintance with sounds, and so on; the acquaintance being the same in each case. 11 We cannot, Price claims, draw on the relation of acquaintance or consciousness to distinguish between different kinds of sensing because the nature of such mental episodes is entirely determined by the nature of the object that we are presented with in the relevant case. The differences between such episodes are "wholly on the side of the data".
The transparency claim that we find in Moore and other sense-datum theorists is thus at the same time both more and less ambitious than the intentionalist reading takes it to be. It is, in one sense, less ambitious because the appeal to introspection is not meant to establish the thesis that our sensory experience involves a relation to the physical environment. The transparency of consciousness, as Moore understands it, leaves entirely open the question about the status of what is given or presented in sensory experience. Indeed, it is important to keep in mind that for the sense-datum theorists the argument from illusion was a central philosophical issue. Unlike current-day intentionalists, the sense-datum theorists were greatly concerned by the possibility that our sensory experiences should be systematically deceptive. Introspection alone is not for them a basis on which to establish the claim that perception involves a relation to mind-independent objects.
In another sense, however, the claim is significantly more ambitious than intentionalists have taken it to be. For Moore's remarks about transparency are not only remarks about the way sensory experience appears to us when we introspect. They are of course that as well. But they are in the first instance a claim about the nature of such experience. Attending to the way that sensory experience appears to us is sufficient, according to Moore, to make a minimal but significant claim about the nature of such experience, namely the claim that such mental episodes must be individuated in terms of an object with which the subject of the experience is presented. This may be contrasted with the claim that intentionalists believe can be found in Moore's writings. That claim serves as a point of departure for an argument the conclusion of which involves a claim about the nature of sensory experience.

| BRINGING SENSORY EXPERIENCE INTO VIEW
In The Varieties of Reference, Evans makes a claim about our capacity to know our own perceptual experiences that is-or, at any rate, at first seems-similar to Moore's claim about the nature of sensory experience. 12 The only way a subject can bring into view how things perceptually seem to him, Evans claims, is by going "through exactly the same procedure as he would go through if he were trying to make a judgment about how it is at this place now, but excluding any knowledge he has of an extraneous kind". 13 Of course, to arrive at a full-fledged self-ascription of how things perceptually seem, the subject cannot just judge how things are. He must also be able to construe his judgment as a judgment about a perceptual experience. Evans argues that this involves having the notion of an appearance: it requires an understanding of the fact that in making a judgment about how things are at a certain place one takes up a certain stance and gives expression to how things-based on experience-seem to one. 14 This part of Evans's account is less important for our purposes and I will leave aside the details here. What we are interested in here is that, according to Evans, the core procedure for describing how things perceptually seem involves "looking outward." The only way of specifying what we perceptually experience, Evans claims, is by focusing on the "things themselves" and judging how they are. Just like for Moore, the specification of a sensory experience is "transparent to" the question of what we are presented with in that experience.
Of course, it would be incorrect to state that, in Moore's case, the question of how things perceptually appear to a subject is transparent to a world-directed question about how things are. As we have seen, his transparency claim leaves entirely open what the status of the objects of experience is. One could say, at most, that the question of how things perceptually seem is transparent to the question of what a subject is perceptually presented with. But the latter question involves no reference to the environment, nor does the subject need to make a judgment in order to answer it.
Indeed, it is a mistake to look for a common insight to Evans's and Moore's claims: once one takes into account their respective argumentative context, the two claims turn out to be orthogonal to one another. (The relevant claims are orthogonal to one another; the conceptions of sensory experience to which they are tied are however, as we will see, at odds with one another.) When Moore makes his remarks about the transparent character of sensory experience, his objective is, as we have seen, to present, inter alia, a claim about the nature of sensory experience. Evans, on the contrary, makes his remarks in the context of an attempt to account for our capacity to self-ascribe mental properties: he is not, in that context, seeking to provide an explanation of the nature of sensory experience.
Questions about the nature of sensory experience are dealt with in another part of The Varieties of Reference, well before the treatment of self-ascriptions of mental properties. Now, the fact that the framework developed previously, in Chapter 5 of the book, should allow to account for self-ascriptions of experiences in the way outlined above is not without significance for the evaluation of that framework itself. As already mentioned in the introduction, and as we will see in detail in Section 4, Evans seeks to reject what he calls a Cartesian conception of the mind-and since, on his view, his explanation of self-ascriptions shows that the framework presented in Chapter 5 can contribute toward accomplishing that task, the analysis of self-ascriptions ends up playing a role in the case he has to make in favor of his general argumentative framework. Yet, the fact that the analysis of selfascriptions should increase the appeal of that framework (by Evans's lights at least) changes nothing about the fact that that analysis is grounded in the aforementioned framework and is not itself a part of it. The claim that selfascriptions of perceptual experience involve a world-directed judgment (and not an "inward glance") is a corollary of a certain conception of perceptual experience and not itself a claim about the nature of such experience.
Here is the central claim that Evans makes about the nature of perceptual experience: In general, we may regard a perceptual experience as an informational state of the subject: it has a certain contentthe world is represented a certain wayand hence it permits of a nonderivative classification as true or false. For an internal state to be so regarded, it must have appropriate connections with behaviourit must have a certain motive force upon the actions of the subject. This motive force can be countermanded, in the case of more sophisticated organisms (concept-exercising and reasoning organisms), by judgments based upon other considerations. In the case of such organisms, the internal states which have a content by virtue of their phylogenetically more ancient connections with the motor system also serve as input to the concept-exercising and reasoning system. judgments are then based upon (reliably caused by) these internal states; when this is the case we can speak of the information being 'accessible' to the subject, and, indeed, of the existence of conscious experience. 15 When we have a perceptual experience-when I, say, feel a cold evening breeze blowing through my hair-, we are afforded a certain awareness. On Evans's picture of such experience, what awareness we are afforded is determined by the informational state that we find ourselves in. What informational state a subject is in obviously depends on interactions between his sensory equipment and the environment. But it is solely the state itself that fixes what perceptual awareness the subject enjoys.
Evans claims further that informational states have a representational content, and it is this thesis that we must focus on to understand his "transparency claim." The experience that a subject is afforded in virtue of being in a certain informational state represents the world a determinate way-a way which the world may, or may not, be. Indeed, a perceptual experience does not only represent things a certain way, but it also represents them as being that way. The subject may mistrust his experience and conclude that things are not as perceptually represented, but his experience does provide him with a reason for thinking that things are that way. I will come back to this second point in a moment; let me first focus on the claim that an informational state has representational content. ZAPERO Evans considers that the representational content of an experience is specifiable without reference to particular objects: he takes it to be specifiable in "object-independent" terms, by an open sentence with one or more variables (with the number of variables corresponding to the number of objects that the state is about). 16 It is thus not necessary, on Evans's conception of perceptual experience, that there should be a tree to be experienced, on a relevant occasion, for one to visually experience a tree. This is something which the ordinary use of "seeing" suggests, since, on that ordinary use, "seeing" is a success verb: one can only see a tree if there is a tree to be seen. On Evans's view, however, the content of a subject's perceptual experience can, potentially, remain unchanged despite that subject being presented with different objects over a span of time, or with no object at all. It is crucial to note, though, that sameness of representational content is not a sufficient condition for identity of informational states. 17 The reason for this is that an informational state has two distinguishable components: its content and its causal antecedents. 18 Indeed, the fact that the causal antecedents play a role in the individuation of such states means that the state as a whole cannot be specified in object-independent terms. Evans makes the point by stating that while informational states cannot be said to represent particular objects, since their representational content is "object-independent," they can be said to be representations of particular objects, much like a photograph can be said to be a representation of a particular object. 19 Now, the idea that an informational state determines a subject's perceptual awareness does not entail that a subject has awareness of the informational state. The informational state fixes what perceptual awareness a subject enjoys by representing things as being a certain determinate way. But it is crucial to note that when a subject is in a certain informational state, he enjoys awareness of the world as being such-and-such. 20 Of course, as already mentioned, if the subject has additional information about how things are, he may distrust his experience. Additional information-this is the "extraneous information" mentioned in the passage above-can lead him to judge that things are not as his experience represents them as being. Yet, even if he does have such additional information, the experience still is one of the world being a certain way. This (alleged) feature of experience is what Evans calls the "beliefindependence of experience." 21 Even for the subject who is "in the know" about the Müller-Lyer diagram, the perceptual experience represents things as different from what he knows things to be. He knows that the lines are of equal length, but his experience represents them as being of unequal lengths (according to Evans-one may dispute this, of course). For the subject, as far as his experience goes, the lines are of unequal length.
It is this representational dimension of perceptual experience, which leads Evans to advance a claim that seems similar to Moore's transparency claim. The fact that perceptual experience is intentional in the way just outlined means that, despite being an internal state, it cannot itself directly become an object of awareness. Such a state is intrinsically of a certain way for things to be. Any attempt to focus directly on that experience will just provide the subject with an awareness of the relevant way for the world to be. We are thus obliged to "look outward," onto the world, to bring into view our perceptual experience. By making a judgment about the world and relying on a certain understanding of what the making of such a judgment involves, we can bring into view, in an indirect way, what cannot be a direct object of awareness.

| LUMINOSITY
Evans claims that his account of our capacity to self-ascribe perceptual experience shows that capacity to be a fallible one. 22 As I mentioned in the introduction, he places much weight on this alleged corollary of his account because he considers that it provides the account with anti-Cartesian credentials. I will return in the next section to the question of whether establishing that claim is indeed enough to provide an account with such credentials. In this section, I want to look at the case he makes for fallibility.
Evans conceives self-ascriptions of perceptual experiences as stances that are answerable to something which is independent from the holding of such a stance. In this respect at least, self-ascriptions are meant to be like judgments: they are answerable to a fact of the matter, which can make them either true or false. Evans stresses this point when discussing the differences between self-ascriptions of belief and self-ascriptions of experience. After having underlined that his aim is to adopt the same general strategy in accounting for both kinds of self-ascriptions, he presents the following asymmetry: In the case [of self-ascriptions of experience], the subject's concentration, as with self-ascription of belief, is on the outside world: how does he, or would he, judge it to be? The cases are different in that in [the case of experience] there is something (namely an internal, informational state of the subject), distinct from his judgment, to which his judgment aims to be faithful. But it is something necessarily approached in the roundabout way I have described. 23 Contrary to self-ascriptions of belief, self-ascriptions of experience are answerable to something wholly separable from the judging by which such self-ascriptions are made. The deployment of our capacity for judgment serves to bring into view something with an independent standing.
By treating self-ascriptions in this way, Evans seeks to make them vulnerable to the kinds of error that are characteristic of judging in general.
The procedure I have described does not produce infallible knowledge of the informational state, for mistakes of the kind that occur when the subject makes judgments about the world can also produce inaccuracies when the same procedure is re-used for this different purpose. For example, consider a case in which a subject sees ten points of light arranged in a circle, but reports that there are eleven points of light arranged in a circle, because he has made a mistake in counting, forgetting where he began. Such a mistake can clearly occur again when the subject re-uses the procedure in order to gain knowledge of his internal state: his report 'I seem to see eleven points of light arranged in a circle' is just wrong. 24 Note that the fallibility of self-ascriptions is meant to derive from the fallibility of our general capacity for judgment. For instance, when perceiving ten points of light arranged in a circle, we may miscount; and if a judgment based on such a miscount is put to the service of a self-ascription, the error will carry over and affect the selfascription.
This, of course, still leaves open the question of whether our capacity for self-ascription is susceptible to the same kinds of mistakes as our ordinary judgments. In the first sentence of the passage quoted above, Evans seems to suggest that it is. But once one looks closer, it becomes evident that this is not so. The crucial difference, as I will show, is the following. When it comes to the environment, indistinguishability is not a warrant of identity. The fact that two environmental situations should be subjectively indistinguishable does not mean that they are identical. This makes possible a particular kind of error: a subject may possess the conceptual capacities necessary to make a correct judgment about a situation and, yet, not be in a position to make a correct judgment. Evans, however, is forced to rule out the possibility of such errors in the realm of self-ascriptions. The reason for this has to do with the nature of the states that those self-ascriptions are answerable to. Two such states cannot be different but entirely indistinguishable to their bearer. In the realm of self-ascriptions, indistinguishability is a warrant of identity.
One could consider that for a stance to be a judgment-the kind of stance that involves truth or falsity-, it must be susceptible to the aforementioned kind of error. If one does, one would have to conclude-if the claim I am about to argue for is correct-that self-ascriptions cannot be judgments at all. But I want to leave this possible line of criticism aside. It would take our discussion in a different direction than the one that we're interested in here. Pursuing that line of thought would involve showing that self-ascriptions of perceptual experience are less of a cognitive achievement than Evans thinks they are. We're interested here in something else, namely in the tension between Evans's account of perceptual experience and his attempt to develop an anti-Cartesian conception of the mind.
My argument here will involve two steps. Before I can show that Evans makes self-ascriptions of perceptual experience immune to a certain kind of mistake, we need to look closer at what exactly such self-ascriptions are answerable to. Evans glosses the outcome of the procedure by which one self-ascribes an experience as "knowledge of the informational state," and this suggests that that procedure is meant to bring into view the entire informational state of a subject. But if one examines the purpose of the procedure more closely, it quickly becomes clear that this cannot be Evans's considered view. 25 As we have seen, the procedure that Evans recommends requires that the subject make a judgment about "how it is at this place now." Such judgments will, presumably, often involve demonstrative expressions, which are used to refer to objects that the subject is being afforded an awareness of; they will be judgments such as: "That individual is staring at me." The recommended procedure will then issue judgments such as: "It seems to me that that individual is staring at me." Now consider a case in which the subject is hallucinating. In such a situation, there is in fact no individual which is staring at the subject. So the judgment is, when taken at face value, not merely false but inadequate or nonsensical: there is no appropriate individual that the judgment can be about.
Yet, Evans surely intends for self-ascriptions to be possible in cases where a subject suffers a hallucination or a perfect illusion. One could, naturally, take a different stance on the matter. One could consider that, in such situations, the subject is in fact (unbeknownst to him) not in a position to make an adequate self-ascription. But on Evans's conception of experience, there is always a determinate way which things are represented as being in experience. Whether or not the subject is hallucinating, there is the way that things seem to be to him. 26 So, Evans clearly intends for self-ascriptions to also be possible in situations in which the subject is hallucinating (and more generally in situations in which, to use his terminology, thoughts are not well-grounded). 27 Indeed, not only should selfascriptions be possible in such situations, but also those self-ascriptions should also potentially be correct-which they could not be if one had to take into account, when evaluating the truth of such ascriptions, the causal antecedents of the relevant informational state. 28 The aim of self-ascription, on Evans's picture, is to capture the experiential character of the relevant experiential episode, and that character is determined by the content of the informational state.
Consider now, more concretely, what a mistaken self-ascription amounts to. Evans claims that perceptual experience has representational content, but he is keen to distinguish between perceptual experience and belief, and to avoid the conclusion that enjoying a perceptual experience just is believing such-and-such. He distinguishes between the two by introducing a much-debated distinction between nonconceptual and conceptual content. 29 Perceptual experience, Evans claims, is a nonconceptual informational state: having such an experience does not require the deployment of any particular conceptual skill. 30 This means, inter alia, that we may fail to have the concepts required to grasp an experience conceptually and may thus be incapable of making it the object of a judgment. One could, of course, raise the question of whether perceptual experience can have truth-evaluable content without possessing the kind of discursive or conceptual character, which thoughts-the paradigmatic truth-bearers-possess. But, for our purposes, we need not take a stand on that matter.
The important point is that a perceptual experience, in virtue of having a certain content, represents things as being a determinate way. The content is truth-evaluable: for every such content there is a conceivable judgment whose content is individuated by precisely that way for things to be. 31 Of course, we may be incapable of making such a judgment because we lack the conceptual resources to articulate the relevant way for things to be. But if judgments are individuated in terms of truth-evaluable content, as perceptual experiences are, there must always be a conceivable judgment corresponding to every experience. More importantly, there must be one conceivable judgment that would capture how things are represented by the relevant experience. So if one possesses the relevant conceptual capacities and is in a position to make a judgment about one's experience, there is exactly one judgment that will capture the content of the experience. There is, in that sense, the correct answer to the question of what perceptual experience one is currently enjoying. 32 Of course, in any given situation there is an indefinite number of mistakes, which can lead the subject to not provide that correct answer. But what determines the correctness of the self-ascription-namely, the truth-evaluable content of the perceptual experience-is something that the subject is entirely aware of. On Evans's account of perceptual experience, enjoying such an experience just is being aware of the relevant content. The subject may not be in a position to make the correct judgment because he lacks the appropriate conceptual resources. But simply in virtue of enjoying the relevant experience, he is entirely aware of what would-were he capable of making it-make a judgment correct. What this entails, as I will now go on to show, is that if the subject possesses the capacities to make a correct judgment about what perceptual experience he is being afforded, he is eo ipso in a position to make such a correct judgment. He may go on to make a mistake, by deploying his capacities inadequately, but he is nonetheless in a position to make a correct judgment-and if he ends up making a mistake that is solely his failure. Most crucially, he cannot make a mistake because the content of his experience has misled him by seeming to be something that it is not. In the sphere which self-ascriptions are answerable to nothing is hidden, everything lies open to the subject's view.
To see how self-ascriptions are immune to a certain kind of error, let us pursue the following issue: does perceptual experience itself admit of perfect illusions? The question here is not simply whether Evans can make room for the possibility that a subject should have perfect illusions, that is, the possibility that perceptual experiences of different ways for something to be should be subjectively indistinguishable. The question is rather whether there can be two different perceptual experiences, which possess different representational content, but which are indistinguishable to the subject. In other words, for a perceptual experience in which things are represented as being F, could there be a different one in which things are not represented as being that way, but where, for the subject enjoying that experience, this would be unrecognizable? Take for instance an individual-call him "Price"-who is afforded visual awareness of a tomato. On Evans's account of perceptual experience, Price is afforded a visual experience with a certain representational content. Now, if the object that Price is presented with were not a tomato but a wax ringer, Price would be suffering a perfect illusion; he would be afforded a false representation of how things are. But could that too turn out to be a false? Could it turn out that, undetectably to the subject, what was represented as so is that there is a tomato or a wax ringer before him?
Before answering the question, let me introduce an important qualification. On Evans's account of perceptual experience, such an experience is an informational state. So if we ask about whether perceptual experiences admit of ringers, our question is about whether informational states admit of ringers. Now, given that the identity of an informational state depends not only on its content, but also on its causal antecedents (it also depends, in Evans's terms, on what a state is a representation of ), it is quite easy to conceive of ringer states. Two states could have the same content but different causal antecedents: they would be indistinguishable to the subject but different in nature (much like qualitatively identical photographs of different objects). But, as we stressed previously, self-ascriptions of perceptual experience seek to bring into view not the entire informational state but the content of such a state. Our question here is thus not simply whether perceptual experiences admit of ringers but whether their content does.
Moreover, it is important to keep in mind that that content is what determines the awareness that the subject is afforded in enjoying an experience. By being in an informational state with a certain content, the subject is afforded awareness of things being represented a certain way, where his having that awareness means that he can recognize just how things are represented. The fact that he has awareness of the content means that, depending on how things turn out to actually be, the subject can recognize whether things are, or not, as represented in his experience.
To see why the idea of a ringer-content is an incoherent one in Evans's framework, we need to consider again the purpose of the notion of an informational state. The appeal to that notion is part of an attempt to explain our cognitive lives in terms of the idea of an informational system; it is meant to help explain why, given the state of his environment, a subject takes to be true what he in fact takes to be true, and why he acts as he in fact does. Perceptual experience is the paradigmatic informational state, and the notion of such a state is meant to explain, in particular, the role that perception plays in a subject's thinking and acting. It is meant to account not only for how a subject can make correct judgments about his perceived environment, but also, and perhaps more importantly, how a subject can be led to make false judgments. 33 Evans seeks to account for both these scenarios in terms of the idea that experiences provide the subject with a reason for taking things to be a given way. The subject may distrust his ZAPERO experience and come to different conclusions, but simply in virtue of being in an informational state, a subject has a reason to take things as being a certain way (this is precisely the "belief-independence" of experience).
The notion of an informational state can only play this explanatory role if the subject is aware of how such a state represents things to be. He must be aware in the sense of being prepared to recognize whether or not things were as represented in his experience. If representation and recognizability come apart, as they would if one allows for ringer-content, the notion of an informational state-and of representational content-could no longer account for what they are meant to account for. They could no longer help explain, as they are meant to, why, given the state of the environment, a subject takes things to be as he in fact does. For the informational state would no longer provide the subject with a reason to take things to be a certain way-it would at least not be a reason that the subject can recognize and that can thereby play a role in the subject's cognitive life. It is thus crucial for Evans that informational states should not be sub-doxastic states: the subject must have access to them in the demanding sense just outlined. Only then can they have the bearing on a subject's cognitive life, which, in Evans's framework, they must have. Now, ruling out the idea of ringers for perceptual experience means ruling out the possibility that the subject should make a certain kind of mistake when self-ascribing such experiences. If there can be no ringers, a subject cannot make a mistake because the content of his experience is such that it seems to be different than it is. The point here has to be framed with care. It can of course seem to the subject that his experience is different from what it actually is. That is one way one may describe the situation in which he commits a mistake. (When he miscounts, it seems to him that he is having an experience of eleven lights, when in fact his experience is one of ten lights.) What cannot occur is that the subject should possess the capacities to make a true judgment-as in the case when he counts the lights-but not be in a position to recognize how his experience represents things as being. This is precisely what ruling out ringer-contents amounts to. So while things can seem to the subject to be different from what they are, the distinction between seeming and being is not applicable to the content of his experience. We may put this by saying that the content is luminous for the subject. 34 The subject may of course lack the conceptual capacities to articulate a correct judgment about his experience.
But the important point for our purposes is that if he possesses those capacities, he is eo ipso always in a position to correctly self-ascribe the relevant experience. Self-ascriptions of experience are thus certainly not incorrigible. Nor is the object of those self-ascriptions evident for the subject, if by evident we mean a condition p such that if p obtains the subject believes that p. But there is still an important sense in which nothing can be hidden from the subject: if he possesses an adequate capacity, a true judgment is always in reach. While the domain to which his self-ascriptions are answerable is not evident for him, it is one in which the mistakes that ringers provide for are ruled out.
Is this news for Evans? One might think that it is not, or at least not important news. We have, after all, allowed for the possibility that the subject should be genuinely mistaken about his own perceptual experiences. And is this not all that Evans lays claim to? Even if fallibility were Evans's sole concern, one might feel uncomfortable about the fact that self-ascriptions of perceptual experience are exempt from a kind of error which plays a central role in our dealings with the environment. But there is a much more immediate problem. For, as I have already pointed out, Evans is not interested in fallibility per se: his concern about the fallibility of self-ascriptions is grounded in his rejection of Cartesianism. And on Evans's own understanding of Cartesianism, rejecting Cartesianism involves rejecting the idea that all mental states are luminous. Or so I will show in the next section.

| INNER VISION
In his explanation of our capacity to refer to ourselves in the privileged way that is associated to the first person pronoun "I," Evans draws heavily on his analysis of demonstrative reference. He considers thoughts involving selfreference to have the "same general character" as thoughts involving demonstrative reference, 35 and he thus uses his account of the latter as a basis to analyze the former. Since perceptual relations play a central role in his account of demonstrative reference, such relations also end up playing a decisive role in his explanation of self-reference: Evans seeks to explain the kind of self-reference that is usually achieved by using the pronoun "I" in terms of the perceptual relations each subject entertains to him-or herself. Yet, it is a central thesis of Evans's account of selfreference that those perceptual relations are relations to a certain human being, to the same empirical creature also identifiable by others. So while Evans accounts for self-reference by appealing to our perceptual capacities, this appeal is not at all an appeal to introspection or "inner sense." On the contrary, in the context of his analysis of self-reference, Evans seeks to flesh out an anti-Cartesian conception of the self. When referring to ourselves with the help of "I," we don't-Evans wants to show-refer to a disembodied mind or ego, but instead to the embodied being which is publicly observable. The central claim is pithily expressed at the very end of the chapter on self-reference: Our thoughts about ourselves are in no way hospitable to Cartesianism. Our customary use of 'I' simply spans the gap between the mental and the physical, and is no more intimately connected with one aspect of our self-conception than the other. 36 The fact that certain judgments about bodily awareness and self-location should be immune to error through misidentification and (allegedly) involve the same kind of object-dependence as demonstrative judgments is meant to show that knowledge of ourselves as physical and spatial beings plays an essential role in our capacity to refer to ourselves. 37 This set of considerations about the self is closely tied to another set of considerations about the mind. For Evans does not just seek to dislodge a Cartesian conception of the self, he also seeks to call into question a certain conception-a conception, which he also dubs "Cartesian"-of the mental aspect of the self. In a sense, this latter set of considerations, targeting a Cartesian conception of the mind, is more fundamental than the considerations about the self: Evans holds that Cartesianism about the mind is one of the central obstacles to embracing the notion of object-dependence which undergirds his anti-Cartesian conception of the self. 38 Indeed, the idea of a Cartesian conception of the mind emerges in The Varieties of Reference well before any issues about the self or subjectivity have been broached. It first emerges in the context of an analysis of Russell's conception of object-dependent thought. The fact that Russell, after having introduced the idea of such a thought, should have proceeded to radically restrict the domain of things that such thought can be about (and, thereby, also radically restricted the very domain of object-dependent thoughts) is understood to be a consequence of a commitment to a Cartesian conception of the mind: It is clear that the restriction Russell imposed upon the epistemological relation which could sustain genuine reference gives that semantical relation application only to a private language, and many philosophers have rightly deemed this absurd. But few have understood Russell's reasons for the restric-

tion. […]
Given Russell's restriction, a situation can never arise in which a subject thinks that he is having or expressing a thought about an object while failing to do so; and this was a possibility which Russell very much wished to rule out, because it seemed to him incoherent. Russell worked with a conception of the mind which was thoroughly Cartesian: it would not make sense for a subject to be in error as to whether or not he satisfied the mental predicate 'ξ is thinking about some thing'. 39 Evans's diagnosis here is grounded in the following line of reasoning. If by "singular thought" we understand a thought the truth of which depends on how a particular object is, a singular thought is object-dependent in the following sense: one can only specify the truth conditions of the thought by referring to the appropriate object. For instance, a singular thought about the suspicious individual that I can presently see at the end of the hallway is an object-dependent one: I can specify when things would be as represented by my thought only by referring to a perceived object, namely the individual that I am presently observing.
It is a trivial consequence of such an idea of object-dependence that there is no singular thought to be entertained when the relevant object-the particular object reference to which was (allegedly) necessary to specify the truth conditions of the thought-does not exist. If we are to credit a subject with the thinking of a singular thought, in the aforementioned sense, it cannot turn out that the relevant object does not exist. If it turns out that what looked like a suspicious individual at the end of the hallway was nothing but a deceptive shadow of a nearby statue, I could not have been entertaining a singular thought about such an individual. There was no such individual to be thought aboutand thus no relevant singular thought to be entertained. Of course, one may want to resist this conclusion. One may argue for instance that even in the case where there is an individual in the hallway, our perception-based singular thought is one about what looks like such an individual-something which is available whether or not the individual exists. That is, one may want to restrict singular thought in a way similar to the way that Russell did, in order to ensure that there cannot be the illusion that one is entertaining a singular thought.
Evans considers that one can (pace Russell, or pace the aforementioned line of thought) hold onto the idea of object-dependent thought without radically restricting the domain of objects about which we can entertain singular thoughts. Singular thought, he claims, can be about objects in our shared environment, so the kind of thing for which inexistence is in the offing. The following argumentative strategy is meant to make this claim defensible: when it turns out that an object does not exist, we find out that we only seemed to be entertaining a singular thought. 40 The discovery that an object does not exist entails simply that we were under a wrong impression about what thought we were in fact entertaining. Perhaps we were entertaining a thought similar to the one we had ascribed ourselves, but we certainly were not thinking the singular thought that we had the impression of having grasped. No such thought was available.
Evans accounts for Russell's reaction to the possibility of inexistence in terms of the "disjunctive" strategy just sketched. Russell rejects the idea that ascriptions of singular thought should be construed as disjunctions (either the subject is actually thinking a singular thought or she is just under the impression that she is thinking such a thought) and, by rejecting this strategy, he is forced to rule out the possibility that things should seem to a subject just as they do when she is entertaining a singular thought, but that, for all that, she should not be entertaining a singular thought because there is no appropriate object. The rejection of the disjunctive strategy forces Russell to restrict the domain of objects about which we can entertain singular thoughts to those which could not possibly not exist. It thus leads him to embrace a conception of thought according to which discoveries about the environment cannot have a bearing on the correctness of our self-ascriptions. Not implausibly, Evans dubs this conception "Cartesian." 41 In an essay in which McDowell further develops Evans's criticism of such a Cartesian conception of thought, he describes that conception in the following terms: [S]hort of the fully Cartesian picture, there is nothing ontologically or epistemologically dramatic about the authority that it is natural to accord to a person about how things seem to him. This authority is consistent with the interpenetration of the inner and the outer, which makes it possible for you to know the layout of my subjectivity better than I do in a certain respect, if you know which of those two disjuncts obtains and I do not. In this framework, the authority that my capacity for "introspective" knowledge secures for me cannot seem to threaten the very possibility of access on your part to the facts within its scope.
In the fully Cartesian picture, by contrast, with the inner realm autonomous, the idea of the subject's authority becomes problematic. When we deny interpenetration between inner and outer, that puts in question the possibility of access to the external world from within subjectivity; correspondingly, it puts in question the possibility of access to the inner realm from outside. 42 In the essay from which this passage has been extracted, McDowell's main target is, at least initially, a certain conception of thought (most crucially, a certain conception of singular thought). Yet, once McDowell has presented what he takes to be the fundamental assumption of such a conception, namely the Cartesian conception of the mind that I have just outlined, he makes an additional step and frames his target in more general terms: he extends his target to perception as well and casts the issue as one about the authority that a person has with respect to how things seem to her. The mark of such a conception of the mind is that it rejects construing self-ascriptions of seemings in the disjunctive terms described above.

Is McDowell being loyal to Evans by extending to perception the criticism of a Cartesian conception of thought?
It would have been open to Evans to refuse any such extension. He could have rejected a Cartesian conception of thought without aiming to defend an analogous position with respect to perceptual experience. That would have involved defending a picture roughly along the following lines. On the one hand, ascriptions of (at least some) thoughts are construed in disjunctive terms: either a subject actually entertains a certain thought or she is only under the impression of doing so. Ascriptions of perceptual experience, on the other hand, are immune to the kind of error that makes such a disjunctive account necessary. Indeed, if by "anti-Cartesian conception of the mind" we merely mean a conception that denies that all mental states satisfy the condition that Cartesianism imposes on such states, then such a position would count as anti-Cartesian. For it does grant that some mental states (i.e. at least some thoughts) admit of a disjunctive analysis. But Evans does seem to want to extend his criticism of Cartesianism to his account of perceptual experience.
We have already seen that Evans is keen to stress that, on his account of self-ascriptions of experience, our capacity to self-ascribe experience is fallible. And, as we have also seen, establishing the fallibility of capacities to self-ascribe mental properties is not an end in itself: it is relevant for Evans as part of an attempt to dislodge a certain conception of the mind. So the emphasis on that feature of his account of self-ascriptions of experience already suggests that Evans wants to extend to perception the criticism of Cartesianism elaborated in the context of his analysis of singular thought. More importantly, when he comes around to presenting his account of self-ascription of experience, Evans stresses a certain parallel between that account and his account of self-ascriptions of belief. The former is meant to call into question a conception of first-person authority that was already targeted by the latter. 43 One could of course raise the question of whether, given his motivations for rejecting Cartesianism, Evans needs to extend to perception his criticism of a Cartesian conception of thought. But, be that as it may, there is evidence that he in fact wishes to do so.
Yet, the fact that Evans may have succeeded in making a case for the fallibility of our capacity to self-ascribe perceptual experience does not warrant the inference that his account is anti-Cartesian according to his own understanding of "anti-Cartesian." The reason for this is as follows. Evans at times characterizes the idea of a Cartesian picture of the mind in terms of the general idea of fallibility. For instance, when he first introduces that picture in the passage quoted above, Evans characterizes it as a picture of the mind for which it makes no sense that a subject should be in error as to whether she satisfies a given mental predicate. Russell, Evans remarks, "worked with a conception of the mind which was thoroughly Cartesian: it would not make sense for a subject to be in error as to whether or not he satisfied the mental predicate 'ξ is thinking about some thing'." 44 It is important, however, not to lose sight of the argumentative context in which the idea of such a conception of the mind emerges. The context is an attempt to understand why one would be tempted to radically restrict the domain of object-dependent, singular thoughts. In the context of this endeavor, Cartesianism is in the first instance a position which rejects a "disjunctive" approach to the ascription of thought. More specifically, a Cartesian conception of the mind is a conception which cannot make room for object-dependent thought about environmental objects because it rejects a certain possibility: it rejects the possibility that things should seem to the subject just as they do when he entertains a singular thought but that, unbeknownst to him, he should not be entertaining such a thought because there is no appropriate object.
The mark of such a conception is that it rules out this kind of mistake-the mistake that "perfect seemings" or ringers can give rise to.
Consider, for a moment, a conception of thought according to which all thought-content is object-independent.
Such a conception is perfectly compatible with a view on which a subject may be mistaken in his self-ascriptions. Fallibility would be guaranteed, for instance, if self-ascriptions had to be made on the basis of a subject's publicly observable behavior. (A subject may be mistaken about what (object-independent) thoughts he may be credited with on the basis of his behavior.) Yet, it would clearly classify as a Cartesian conception of thought since it would rule out the kind of thought that a rejection of Cartesianism is meant to make room for. Evans makes room for objectdependent thought by introducing the idea of a "disjunctive" approach to the attributions of thought-and that strategy involves introducing more than just fallibility about one's own cognitive life; it involves introducing the error characteristic of ringers-so errors which the subject, on the relevant occasion, is not in a position to correct.
Take for instance the thoughts about two qualitatively identical Doppelgänger that a subject may entertain on the basis of similar perceptual experiences (or perhaps even perceptual experiences with identical content). Rejecting Cartesianism is meant to allow us to say that, in the two cases, the subject can, unbeknownst to him, entertain two different thoughts: thoughts the truth of which depends on two different, particular individuals. What this requires, however, is that one allow for the possibility that a subject should not be in a position, even if he has the relevant discriminatory capacities, to distinguish between two such thoughts. Allowing for the possibility of such errors is what makes a conception of thought anti-Cartesian. In other words, the mark of anti-Cartesianism is in the first instance the rejection of luminosity.
Evans's attempt to make room for the idea that self-ascriptions of experience are fallible therefore comes too late. It is a mistake to think that such an attempt could, even if one deems it successful, provide his account of such self-ascriptions with anti-Cartesian credentials (again, by Evans's own lights). For once his variety of intententionalism is in place, the central tenet of a Cartesian outlook has been accepted: the luminosity of perceptual content is in place. Any subsequent attempts to allow for fallibility are inevitably insufficient. Self-ascriptions of perceptual experience may be fallible, but they are answerable to the content of an experience, and that content is luminous. The key is that Evans rules out from the outset an option that Moore's transparency claim still leaves room for. In claiming our perceptual experience to be transparent, Moore makes the nature of a perceptual experience dependent on the existence of an object. Of course, he goes on to claim that the relevant objects are not the objects of our environment-and his general account of perceptual experience may thus seem prototypically Cartesian. But the transparency claim in itself does not yet bear on that matter. It leaves room for the idea that a perceptual experience should only be specifiable with reference to objects of our shared environmentand thereby also leaves room for the idea that discoveries about the environment should have an impact on ascriptions of such experience. But this is precisely what Evans's account of perceptual experience cannot allow for. 45

David Zapero
https://orcid.org/0000-0001-5342-5414 ENDNOTES 31 Someone committed to the analysis sketched in Footnote 27 could deny that all perceptual experiences have truthevaluable content. If the content of a hallucinatory experience were articulable in judgments only in the scope of make-believe, that content would not be truth-evaluable. I have already expressed my reservations about such a position. It is crucial to keep in mind the difference between judging that something is make-believedly the case and makebelievedly judging that something is the case. Put in terms of Evans's sentential operator "*(x)*": we must distinguish between judging that *p* and *judging that p*. The aforementioned proposal consists in construing self-ascriptions of hallucinatory experience as acts whereby a subject *self-ascribes that things seem such and such*. (We want to end up with: *the subject's self-ascription is true* if and only if *things seem to her as she purports*, and not: the subject's self-ascription is true if and only if *things seem to her as she purports*.) But this means sacrificing the idea that, in the case of hallucinatory experience, self-ascriptions are possible-all it leaves us with are make-believe self-ascriptions. As I have already pointed out, Evans does seem firmly committed to the possibility of a subject being in a position to self-ascribe such experiences.
Note also that such a proposal goes against Evans's fairly categorical assertion (Evans 1982, p. 226) that perceptual experiences have truth-evaluable content. 32 I have avoided dealing with the conceptual/nonconceptual distinction, but, of course, my argument does bear on that issue.
The line of reasoning that I have presented is a prominent way of attacking the very idea of nonconceptual content. Of course, one can adopt such a general strategy-show that any determinate truth-evaluable content must at least potentially be the content of a judgment-and draw quite different conclusions. One may adopt that strategy, reject the idea of nonconceptual content and then conclude that perceptual experiences have no truth-evaluable content at all; or one may adopt such a strategy, reject the idea of nonconceptual content and conclude that perceptual experiences have conceptual content. For the former argument, see Travis (2009, pp. 156-163); for the latter argument, see McDowell (1998a) and McDowell (1998b). 33 Consider how Evans (1982, pp. 123-124) stresses the significance of belief-independence when introducing the notion of an informational state. 34 My use of this term is similar but not identical to the use that Williamson (1996) makes of the term. On Williamson's use, a condition is luminous simply if a subject is always in a position to know whether the condition obtains (the factivity of knowledge allows one to reformulate the definition as a bi-conditional). On my use, luminosity is something broader. A condition is luminous if it satisfies the following constraint: if a subject possesses the capacities to make a correct judgment about whether the condition obtains, he is eo ipso always in a position to make a correct judgment about that condition. (In fact, what I have called luminosity is very close to what Williamson (1995, p. 535) calls "transparency," but for obvious reasons I have avoided using that term (again).) The luminosity I am interested in is most explicitly dealt with in discussions about so-called observational predicates. Such a predicate can be defined as a concept which satisfies the following condition: it is inconceivable that an observer, through the exercise of certain perceptual capacities p 1 , p 2 , and so on. should take an item to be F and, where no malfunctioning of those capacities occurs, that that item should not be F. For a discussion of the very idea of such a predicate, cf., most notably: Dummett (1975), Travis (1985), Wright (1975). 35 Evans (1982, p. 205