Seeing is, or affords, a certain sort of awareness – visual – of one's surroundings. The obvious strategy for saying what one sees, or what would count as seeing something would be to ask what sort of sensitivity to one's surroundings – e.g. the pig before me – would so qualify. Alas, for more than three centuries – at least from Descartes to VE day – it was not so. Philosophers were moved by arguments, rarely stated which concluded that one could not, or never did, see what was before his eyes. So much for the obvious strategy. It occurred to almost no one to object that this could not be right. Frege did, but no one noticed. Austin, finally, did away with that conception of good faith in philosophy which had allowed such a thing to pass, and then with those arguments themselves. Until then, philosophy was deformed. Robbed of the obvious approach, a Drang set in to gaze inward, hoping to find what it really is to see in what enabled sensitivity to pigs, or in its byproducts. Gazing inward can be science, but often merely poses as it. It can be difficult to disentangle actual science (or at least empirical fact) from mere preconception pretending to its rigour. Most nowadays feel rid of the grip of those barriers to the obvious approach. But, as we shall see, many so feel wrongly. The Drang still misshapes their thought. I aim here to identify the Drang at work; thereby, I hope, to rid us of it.

1. Seeing

One sees, one would have thought, such things as mangoes, sloths, the setting sun, the sun setting, the blackened condition of the toast. An account which said otherwise would, one would have thought, thereby be shown mistaken. As noted, for more than three centuries the nearly universal, and unquestioned, view was: ‘Not so.’ What one really saw, it was supposed without a blush, were, in Frege's terms (pointing to the solecism here) contents of one's consciousness: objects of awareness such that, first, there is someone one would need to be to be aware of them, and, second, their career as something to be aware of was coeval with that person's awareness of them. One saw things in one's surroundings, if at all, only insofar as seeing these other things might pass for seeing that.

H. A. Prichard – an otherwise admirable philosopher – exemplifies the frame of mind. He begins a pre-war essay, ‘Perception’, by admitting that in the ‘everyday attitude of mind’ of philosophers and others, one counts ‘“chairs and tables, boats going downstream”, and so forth’ as the sorts of things one sees and touchesFootnote ¹ then commenting,

It will not stand examination because of a principle which Prichard expresses in the following ways:**

Given the principle, the idea that we see such things as boats, or walls, is refuted by such ordinary facts as that the moon (roughly spherical) sometimes looks flat, or that if you look at a wall in a mirror left and right reverse.

Many of us have, at one time or another, failed to see a wall (in time). But, one might think, if a principle entails that one can never see a wall, then, in point of good faith, it is, ceteris paribus, false. Prichard offers a reason for thinking otherwise. He reminds us that we are ‘concerned simply with the nature of what we see in the proper ‘mental’ sense of see’ and not ‘with what we see in the physical sense of what affects our eyes.’Footnote ⁴ So our idea that we see such things as sloths and walls comes from our thinking of some ‘physical sense’ of ‘see’, whereas Prichard, and any philosopher of perception, means to speak of another. It is not clear what either of these senses is; nor even that there are different senses of ‘see’.

All the same, Prichard's talk of physical and mental seeing is suggestive. No one, I think, thinks there is any sense of ‘see’ in which merely to have something ‘affect your eyes’ – say, form retinal images – is to see it. Why not? A natural, though not inevitable, idea is: before one saw anything, more stuff would have to happen. (Not inevitable: to have images on one's retinas would not be to see something whether or not more stuff needed to happen.) A further natural idea might then be: at a certain point the relevant stuff, or enough of it, has happened. At that point, the perceiver goes into a certain ‘internal’ state, the upshot of the stuff, where this is one of a specified range of states into which a particular device, ‘that which enables vision’, might go. In such a state, the idea is, one enjoys visual awareness (or experience). One sees only in enjoying visual awareness. The state decides what visually awareness one thus enjoys; thus, at least, what it is in which one might be seeing something.

The nature of the state is to be fixed (causally) by that of which it is an upshot. Any retinal state might unleash any of various sequences of stuff happening. Retinal states thus underdetermine what such internal state one arrives in. So, accordingly, for whatever one has in view; whatever formed images on one's retinas. Viewing that might place one in any of various internal states, each differing from the others in what one is thereby visually aware of. Conversely, for any such terminal internal state one, what was actually in view before one might be any of many things – perhaps even nothing.

A crucial idea here is that such an internal state furnishes all the visual awareness one enjoys. It thereby furnishes an answer to the question just what one is then visually aware of. So, for one to be in it is for there to be a determinate way one experiences things being (or seeming) visually. For there to be a determinate way, as this is usually understood, is for such-and-such to be the way; thus for one to experience things being, or seeming, visually (arranged) thus and so. Perhaps being visually aware of what it thus decides you are, while, in fact, viewing given things – say, a pig – is what it is to see (so be visually aware of) something in one's surroundings – say, the pig. If so, then that much as to what one is visually aware of, the internal state would not determine. Modulo that possibility, the internal state you are in at the time answers the question what you are then visually aware of.

I will call the whole of the above the SH model (‘SH’ for ‘stuff happens’.) The mere idea that we are enabled to see by some identifiable states, the products of happenings unleashed by retinal images, may be science, or something it presupposes. The rest is not. It is a picture of what awareness must be. I will next begin to trace the current career of the SH model. I hope to show a few reasons why that picture will not do.

2. Perceptual Experience

Gareth Evans is concerned with visual experience, a wider notion than seeing; still more widely, with perceptual experience in general. He writes, ‘In general we may regard a perceptual experience as an informational state of the subject …’Footnote ⁵ – as he insists repeatedly, an internal state. Such is an odd start. An experience, where one can speak of one, is, one would have thought, an episode, in which one experiences something or other there is to experience – in the case of perceptual experience, typically, though perhaps not always, something impinging from without – the warm summer breezes, perhaps, the patterns of sunlight on the wall. To regard all that as an internal state is, for one thing, to leave what is experienced out of the picture altogether. If experience is to be an internal state, the SH model provides ones for this to be. Could Evans' talk of experiences as states of a subject be symptomatic (or more) of allegiance to that model?

The informational states in question are not themselves objects of their owner's perceptual awareness. In experiencing, say, the barnyard, I am (visually) aware of that pig in the sty; not of any internal state of mine. But, Evans tells us, I can be aware of features of such a state, specifically of the information it contains. For experiencing the barnyard is experiencing being in the state: being in it, I experience. One can, Evans tells us, exploit this way of experiencing the state ‘by re-using precisely those skills of conceptualisation that he uses to make judgements about the world.’Footnote ⁶

So, e.g. it seems to me as though that pig is spotted; ceteris paribus the relevant state contains the information that that pig is spotted. I thus can be aware of the state containing information it does. Just so, for Evans, can perception perform its most central task in a thinker's life: allowing how things are to bear, according to their bearing, on how to think they are; thereby on how to act.

A state which thus contains information is also liable to contain misinformation. It may seem to me as though that pig is spotted when it is not. (Just the shadows of the branches of that spreading chestnut.) One could appropriate ‘information’ to include that. It will avoid confusion not to. Information, here, will be of what is so.

But there are two uses of ‘seems’. On one it seems as though Sid did it if, based on how we see things are, such is most probable. On the other, that pig, standing under those branches, in the filtered sunlight, looks as though it were spotted. It looks (just, or rather) like a spotted pig. It may so look even if we know it is just the light and shadow. It would be a further claim that it does not just seem that way, but seems to be so.

These two readings of ‘seem’ point to two different ways for something to contain information. The scene before me may, in the first way, contain the information that there is a bísaro before me: it makes this recognisable (to one who knows his bísaros). It may contain the information that this pig has swine flu: a decent veterinarian could tell that at sight. That road sign contains information in the second way. It contains the information that Santiago is 48 km. hence: it says so. The information contained in this second way consists of just those things which what contains it represents, truly, as so.

The scene before me, in being as it is, instances literally countless ways there are for a scene, and thus countless ways for things, to be. It is recognisably countless such ways. So any scene contains indefinitely much information in the first way. Which is not a way of containing misinformation: information contained in this way is available through the exercise of recognition capacities; a capacity being, per se, one to get things right. Thus it is that scenes – signposts and billboards aside – are not liable to contain misinformation. If there is something in perceptual experience that is so liable, it will have to be something other than the scene in view, or its inhabitants, again signposts and the like aside. There will have to be something about it, not present in a scene, which allows it to do so.

By contrast, the second way of containing information is by representing something as so. The signpost contains just that information (misinformation) which it represents truly (falsely) to be so. That something was represented as so never means as such that it is so. There is always that much room for representing falsely. So this second way of containing information is a way of containing misinformation.

Frege shows why, if perception is going to make the world bear, for us, on what to think, then it had better do so in affording us awareness of what contains information in the first of these two ways. It had better afford opportunity for exercising, on what we are thus aware of, capacities for thus extracting information. For, to begin with,

In Frege's working of the point, concepts are satisfied, or not, by objects – items not themselves eligible for being satisfied, or not, by anything. A given concept ‘reaches to’ a range of objects which are those things which satisfy it (and identifies a range which are those things which do not). Relations purely between some concepts and others cannot on their own tell us to what range of cases a given concept reaches (though, given to which ranges certain concepts reach, relations internal to the domain of concepts might tell us to what some further one reaches). To grasp a concept is to grasp both how it relates to other concepts, and – a different matter – how it relates to what are not concepts – in Frege's working of the point, to objects: what satisfy, but are not, themselves, satisfied. One must be able to tell, of such things, when they would be such as to satisfy the concept. Such capacities are just the sort one would exercise in extracting information contained in something in the first way. Without such capacities, one would have no concepts, so would not be a thinker at all. There would be nothing on which the world could bear for you. So the possession and exercise of such capacities must be fundamental in any story of how perception makes this bear on what to think.

The point is worth reworking. Frege also says,

The thought is of things being a certain way. The generality of a thought, in the present sense, consists in the fact that there are various ways things being as they are might instance things being that way. If the thought is that that pig is a bísaro, things might be that way while the pig is in a sty, or in open range, while it eats acorns or yams, in a good or mediocre vintage year, and so on. The thought reaches to a range of cases in which things being as they were would be their being that way. Doing so is intrinsic to being a thought at all. What it reaches to is the particular case, which is: things being as they are. Such a thing has no reach to anything.

The generality of a thought is shared by a way for things to be, a way for something to be, and by whatever would make a thought about a way for something to be (call that a concept). It is lacked by what such (first-order) things reach to. I will call what has it (the) conceptual, and what lacks it (the) nonconceptual. The point is now: one is not engaging in thought at all; one is certainly not in a position in which there is any way perception could make the world bear for him on what to think, unless he has, and exercises, capacities to see how the conceptual reaches to the nonconceptual: when things being as they are would be things being thus and so; something's being as it is it being thus and so. Without such capacities for extracting information in the first way, one would not be a thinker at all.

This idea appears in Frege's view of seeing that. Here are two expressions of it:

Sight affords visual awareness of a scene, and of things, happenings, and conditions obtaining in it: that sloth, the waving of the branches, the blackness of the toast. It thus affords opportunity for exercising certain sorts of capacities: I can recognise, of the scene being visibly as it is, that that is that sloth sleeping, or that toast being burnt. Such is a a fundamentally important way for perceptual experience to make the world bear, for me, on what to think.

I return to Evans. In what way does he think one of his internal states contains (retrievable) information? Is it in the first way, so that it would be the pig's looking spotted (to me) which is necessarily closely connected with the state containing the information that it is spotted, or in the second, so that it is my thinking (in the right way) that the pig is spotted which is thus connected with the state containing the information that it is? The first idea excludes misinformation. Which would exclude representing as so. Evans is attached to the idea of misinformation (what an internal state would be intrinsically liable to provide); but also to the idea that perception is occasion for exercising capacities of conceptualisation; thus ones belonging to the first idea. So he is ambivalent. On the one hand, he insists,

Which looks, so far, like the first idea. But on the other, he insists that a perceptual state

Perceptual states are truth-evaluable. Only containing information in the second way could make something so evaluable. Only the second way makes room for misinformation – something representation is intrinsically liable to provide. In fact, Evans seems to want the best (for him) of both worlds: representation, but representation with nonconceptual content. Representation necessarily reaches beyond the particular case which it represents as a certain way. So it belongs to the conceptual. If we draw a conceptual-nonconceptual distinction as above, ‘nonconceptual representational content’ is senseless. Perhaps Evans understands these notions in some other way. I do not so. I think he simply fails to choose where one must. I will return to that presently. First, how does Evans' story reflect the Drang in general, and the SH model in particular?

3. The Inner

Visual experiences, for Evans, are certain inner states. These float free of what lies within the subject's view. For all that in that scene there is a pig beneath the oak, that inner state which is the subject's experience of this may or may not contain the information that this is so. For all that it contains the (mis)information that there is a pig beneath the oak, there may or may not be one. If the visual experience is one of seeing a scene, then that inner state it is is, presumably, the upshot of interactions with the environment; presumably involving, somehow, retinas. So Evans' account just is the SH model. The first thing to note is that it is a necessary condition for visual experience to represent things as so; to be truth-evaluable. For that, experiencing visually must contain something floating free of the scene in view, as per above. There is nothing in visual experience which could do that except, if it exists, something the subject puts there; something inner, independent, in principle, of what surrounds him.

Scenes do not represent things as so. Nor do their constituents. There is no way things are according to that pig before the oak (unless it talks). There is no way things are according to its being beneath the oak, or according to the spreading of those branches. None of these things is truth-evaluable. A pig is not. Nothing's being thus and so is. Correlatively, neither the scene nor the pig nor the spreading contains misinformation. Perceptual experience would need some other ingredient for it to represent anything as so. The spreading of those branches no doubt means various things. Perhaps it tells us something about the average hours of sunlight per year in that place. But there is no such thing as it telling us this falsely. Which highlights the point that such meaning (the factive sort) is not representing anything as so.Footnote ¹³

There are more notions of representing than are found in representing as so. The state of Sid's liver represents years of hard drinking; the state of those rocks aeons of water erosion. Sid's liver thus contains the information that he indulged in years of hard drinking, the rocks the information that this was once a riverbed. Which makes neither Sid's liver, nor its state, truth-evaluable. These do not represent years of hard drinking if there were none. The spreading branches may represent something – say, careful pruning – in this sense. But something else is needed before anything in an experience of seeing them could represent anything as so: commitment. Something in the experience must make itself hostage to whether something is some given way. (Hostage, that is, for correctness. Which might mean harbouring ambitions – as judging is a posture one aims to hold only in a particular sort of world.) The spreading of the branches is not thus hostage to their being pruned; if they were not, it simply does not mean they were.

Representationism – the view that experience represents as so – thus requires the SH model. Does the model otherwise invite it? Adepts of it have tended to see visual experience as representing in some sense. But not always as representing things to be so. Descartes is a good representative of the alternative. For him, all ‘cogitationes’ (ideas) represent. But he also holds,

A room in the Louvre is filled with sycophantic paintings, by Rubens, of Cathérine de Medicis, e.g. astride a horse, in battle armour, leading grateful troops. That painting represents her so doing. Basing one's judgement on the painting, you could take it that Cathérine once, on horseback, led troops. That would be a mistake, which the painting, rightly understood, in no way encourages. The painting does not represent it as so that Cathérine ever did such a thing. The mistake would be all yours. So it is, for Descartes, with perceptual experience in general. My visual experience, as I view the scene, represents a pig in the (or a) sty. I may take it that there is a pig in the sty. But in this I am on my own. No such thing is so according to the experience. It does not represent it as so that any pig is in any sty. Nothing, on Descartes' view, is so according to an experience as such. (Thus far Descartes and Austin are one.Footnote ¹⁶

An Evansian inner state, where one saw a pig in the sty, would represent a pig in the sty in the way that the state of Sid's liver represents hard drinking. It could not represent in that Rubensesque way above. For that it would need something which functioned as a canvas does: (some of) its visual features, themselves objects of the subject's visual awareness, would carry representational import, thanks to something paralleling an intention that they should be taken in a certain way. Evans explicitly rejects any such thing. Might these states still represent things as so? What function, within the SH model, would such representing serve?

Perception's essential task in the life of a thinker is to allow the world – specifically his surroundings – to bear for him on what he is to think (and do) according as it bears on what is so. If cogitationes were, as it were, a mere play of shape and colour, they would play no such role; would make for no such bearing. For that, they must, for a start, carry information as to how those surroundings are. There are two ways for an object of (perceptual) awareness to carry information. It (our its presence) may mean, factively, or indicate something; or it may represent things as some way (on some notion of representing as). Suppose that the play of shape and colour meant things (as well it might if there is such a play at all). A particular visual arrangement in the play might mean, say, that there is a pig before me. Then, if I appreciate its meaning, I can conclude from what I am visually aware of that there is a pig before me.

Frege's point about the conceptual-nonconceptual distinction is one reason why factive meaning, so far as it goes, cannot allow perception to do its job. Some display of shape and colour may mean there is a pig before me (my porcine warning system), just as some grunting sound, or Aunt Ida's shrieks, might. If I appreciate its meaning, any of these, if taken in, may be reason for me to think there is a pig before me. But that of the world which in fact bears on whether there is a pig before me is, in first instance, things before me being as they are. It is that which is, or is not, a pig being before me. Perception's job is not done unless it makes available to me (at least sometimes) that for me then to evaluate, or recognise, as a pig being before me or not. Perception must afford me opportunities to exercise my abilities to link the nonconceptual to the conceptual as it in fact links, by that fundamental relation of falling under, or instance, or just being, as per above. If I am cut off from such possibilities, then, as per above, I cannot so much as get the right bits of the conceptual in mind – a pig being before me, say. Then factive meaning, for all that it may still be there in the world, is no use to me when it comes to what I am to think. To appreciate it, I would need to recognise the bearing of something I do experience on … what? Anyway, on something of which, so far, nothing furnishes me awareness.

The only alternative, then, is to take cogitationes to provide information by, or in, representing things. Descartes wisely resists supposing cogitationes to represent things to be so. So they represent, roughly, Rubensesquely. But this clearly makes no progress. I will put this by harping on a point. A cogitatione is the content of someone's consciousness, in Frege's sense. The scene before me, and its contents – that pig staring at me from the sty – are things for one to experience, to witness, to observe, where, crucially, there is no one one must be to be the one. It is precisely not a content of consciousness. As Frege insists,Footnote ¹⁷ it would be a solecism to speak of seeing a cogitatione, precisely not solecistic to speak of seeing the pig, or the scene. So my cogitationes of the moment being as they are is one thing; the scene before me being as it is another. Accordingly, my being aware of my cogitationes being as they are is one thing; my being aware of the scene being as it is another. Just maybe, I could be (visually) aware of a scene being as it is in being (visually) aware of my cogitationes being as they are (though I doubt it). But awareness of my cogitationes being as they are is not awareness of the scene being as it is. My cogitationes being as they are may be (though Frege, I think, showed not) their being thus and so – e.g. their representing a scene with a pig front and centre. My awareness of their being as they are may make this recognisable to me (if the notion recognition (Erkennung) fits here). But this is not what representation needs to deliver here. What is needed is that my visual awareness of the scene being as it is can make it recognisable to me that that is a pig being before me. No such thing is in the cards. Representation is hopeless for its appointed role here.

Representing makes no progress over factive meaning here. Neither offers awareness of what would be, so could be recognised, as things (before us) being thus and so. Such should have been obvious from the moment cogitationes came on the scene. Whatever they do, awareness of them doing it could not be awareness of a scene doing its thing. But Evans will have no truck with cogitationes. For them to represent would be for their visual features to have representational significance. To take in their representing what they do would be to take in those visual features and appreciate their representational significance. (Anyway, as Frege stressed, such things would need (very roughly) an intention attaching to them for them to represent. Where would that come from?Footnote ¹⁸) Whereas Evans explicitly denies that we are aware how experience represents by being aware of some vehicle whose visible features bear representational significance. Those internal states which represent, for him, are not objects of perceptual awareness. Which leaves him with nothing that could represent Rubensesquely. Which is fine with him, since, anyway, in his view, doing the job right – making the world bear as perception must – requires representing things as so.

So is the representing Evans posits any more fit for its task than Descartes'? Descartes is driven to posit representing because, given what he thinks perceiving is, it is the only hope for perception to give the world bearing on what we are to think. Once cogitationes are in the picture, they must somehow carry information about the world; nothing else could make them do so. Evans' hand is similarly forced. Not that he posits sensory vehicles of representing. But he holds a version of the SH model. On it, an internal state decides what I am visually aware of in experiencing visually: something I might be aware of no matter what the scene before my eyes (visibly) provided. Awareness of that must, somehow, make the scene's being as it is bear on what I am to think – allow me so much as to think, and then to judge, of the scene's being as it (visibly) is that that is a scene being thus and so. What else but representing to connect me with that of which I am thus to judge?

Now the same considerations that defeated cogitationes for this purpose do in as well what such an inner state makes one visually aware of. For awareness of that is not, and could not be, awareness of the scene before me being as it is. (It is another matter whether in visual awareness of the one sort I might sometimes enjoy some awareness of the other.) I am not so far offered so much as the chance to get in mind that which I must judge to be thus and so. All remains just as for Descartes.

If anything, Evans is in a worse position. For representing as so mismatches the conceptual with the nonconceptual. It provides, at best, only bits of the conceptual to bear on other bits of the conceptual; where what is needed is the nonconceptual – the scene before me being as it is – to bear for me on what I am to think as to how various bits of the conceptual reach to it. Experience, on Evans' view, makes available to me, to bear on how I should think of things, at best the fact that the pig is in the sty, where what I need is acquaintance with things being as they are, so that this may bear on whether to think that such is the (or a) pig being in a sty.

Some have thought that experience, to do its job, must supply one with things shaped like (truth-evaluable) thoughts. Christopher Peacocke, for example, tells us,

Only what was shapes as a proposition could bear on the truth of a proposition. Such reflects one conception of rational bearing. But Frege shows how so conceiving bearing gets things exactly backwards, at least when it comes to performing perception's task. Experience must provide that on which I can exercise my capacities to recognise connections between the conceptual and the nonconceptual. Only then can it provide me with that by which I can judge, non-inferentially, that, say, a pig is in the sty. Representing things as so is hopeless for that task. Leibniz on being (so thinking about) an individual is à propos: for no proper subset of an individual's properties (closed under strict entailment) is having them being it.

Experience representing things as so could not be what places us in contact with the world. Such representing would present our surroundings to us as falling under a certain generality. It would thus reach to a range of cases of surroundings being as they were, or a scene as it was – just those instancing the relevant generality. Thus would it represent things as being a certain way. To grasp such representing we would need to grasp when what does (or fails to do) such instancing would does so or not; so get in mind its doing so. If our only route to getting in mind the particular things which do such instancing is through having it represented to us that they do, or do not, instance this or that, how could we so much as think such thoughts? In which case, what would make these the ways our experience represents things to be? (Frege saw the point.Footnote ²⁰ John McDowell made it wellFootnote ²¹) The SH model cuts us off from the world in our visual experiencing-as we must be cut off for representation by, or in, experience to so much as gain a foothold. Once so cut off from the world, it is a nice question how we are ever to be able again to get in touch with it.

4. Nonconceptual Content

So the SH model ends in an impasse. Such are the wages, for perception, of an inward gaze. It can be understandably tempting to fasten on representation for effecting an escape. But it is hopeless for that task. The idea of representing in or by experience – at least as means for experience to bring the world to bear on its subject's thought – positively requires something with just those features of the SH model which condemn it to failure – reason enough to give up on that idea. But it can be hard to abandon the SH model, especially when it masquerades as mere science.

There is, inter alia, this tempting picture. On Evans' account (as on the SH model), in a visual experience there is a way things appear, or seem, visually, to us. Call that, if you like, things appearing, or seeming, a certain way (namely, the way they do). It is just this (through the procedure Evans describes) which makes how our experience represents things recognisable to us. So we are in contact with something belonging to the nonconceptual (things appearing as they do), which we can thus recognise as falling under various generalities – instancing appearing as though a pig were spotted, say. Why cannot that opportunity for exercising conceptual capacities on the nonconceptual be all that is needed for the world's being as it is to be made to bear on what we are to think? After all, the right sorts of capacities are in the picture somehow. Well, it cannot because what is needed is opportunity for exercise of those capacities on the nonconceptual our surroundings supply (or bring into view). But, conflating awareness of A in awareness of B with awareness of B just being awareness of A, that point goes missing.

The very pull of such manoeuvres shows the tension between the SH model's untenability and seeming inevitability. Such tension is, perhaps, reflected in the strange idea of nonconceptual content. Here content is to be the content of what is truth-evaluable; what represents as so (or, again, potential components thereof – what may be true of something). The content of a given item fixes, or is fixed by, what is so according to it; on what its truth turns. To represent as so is to represent things as a certain way. For (such a) representation to have nonconceptual content would be for there to be a way for things to be which was (or was in part) nonconceptual. If we draw the conceptual-nonconceptual distinction as it has been drawn here, this is patent nonsense. A way for things to be, so, too, representation (of things as a certain way) – in Frege's terms, a thought – is essentially what reaches to the nonconceptual in a particular way so as for a thought to present the particular case (things being as they are) as falling under some particular generality. Representational content belongs essentially to the conceptual, on this deployment of terms.

Evans, and his intellectual heir, Christopher Peacocke, are two among many who think that the representational content of experience is ‘nonconceptual’ (or, in Evans, nonconceptualised). Peacocke (soon to be centre stage) begins his treatment of perception with the remark (for him mere truism) that ‘a perceptual experience represents the world as being a certain way.’Footnote ²² That way, he further holds, may be in whole or part ‘nonconceptual’. If they are not just talking nonsense, they must mean something else by ‘nonconceptual’ than the use it has been given here. It can illuminate distortions in the SH model to ask what this can be. (This may oversimplify. Perhaps Evans, or Peacocke, simply gives in, simultaneously, to the pull of both poles of that above tension.)

It might seem less pressing to say what ‘nonconceptual content’ might mean if it seemed absolutely mandatory that experience represent as so. If, for one or another reason, the content of such representation could not be straightforwardly conceptual, well, we can leave it to later to work out just what else it might be. Pro tem ‘nonconceptual’ will do as a placeholder. But, we have seen, it is a mistake (or two) to think any such thing mandatory. So what might Peacocke mean by nonconceptual? It is none too easy to say. ‘Conceptual content’, he tells us, ‘is content of a kind that can be the content of judgment and belief’.Footnote ²³ If we suppose that nonconceptual content is content which is not conceptual, it follows that nonconceptual content is of a sort which cannot be the content of a judgement. I take this to mean: either of things (catholic reading), or of some thing(s) that they are (it is) thus and so. (So that the concept being a crisp is not nonconceptual merely because one cannot ‘judge that a crisp’. But, for both Evans and Peacocke, an experience was to represent things as a certain way; where that is to be read as a way there is for things to be. Just so does Peacocke hope to finesse the ‘problem’ of how experience can provide something which bears on what to think, from which we may learn that such-and-such. Experience represents things as thus and so (on their story); the subject then takes experience's word for it – acquiesces in the representing, so judges – or, if ornery, or suspicious, resists – declines so to judge. On Peacocke's own account of conceptual content, it seems, there could be no nonconceptual content.

In such binds one might try relativising. What, for Peacocke, points towards nonconceptual content suggests he has some such thing in mind. For example, he says,

We, perhaps, could just things to be as represented – just that way. But other creatures, whose experience could still represent things as that way, could not so judge. So goes the thought. If my experience represents Sid just to have grunted, that is a way I, but not these other creatures, might judge things to be. But in this case (the idea seems to be) those other creatures could not so much as have experiences which represented things as that way. For that, they would need the concept grunt. (I blush at such thoughts. But such are issues for other occasions. Anyway, I suppose something like this must be what Peacocke thinks.) So, the thought is, conceptual content would be content an experience could not have unless its subject had the relevant concepts; nonconceptual content is what could be the content of experiences both of those who could judge things that way and of those who could not.

Nonconceptual content in this sense would still be conceptual in the sense of having that generality which, for Frege, is the mark of a thought. It would reach in a particular way to a range of cases, so as to present the particular case – what is experienced – as falling under a certain generality. Only then could it so much as be something we could judge. The main question, to follow, is how any experience could arrange for anything to do that. But, again, necessity is the mother of bracketing such issues. This done, Peacocke's idea, above, is, I think, just another manifestation of the original tension: representation must, but cannot, make the SH model work. I will call those other creatures, who cannot judge what we can, ‘cats’. Why think their experience has this content, which they cannot recognise, or so much as take it to have (on pain of being able to judge so)? The idea is: (details of cat eyes, etc. aside), if I am looking at a sparrow in a bush, and a cat is looking at a sparrow in a bush, we might see, and experience visually, the same thing. It is not as if the cat must be blind to some region of the scene where I can see what is going on. But representational content, if there is to be some, must be a function of the character of the experience – what it was like (visually). Accordingly, that sameness in what I and the cat experience visually must be reflected in a shared representational content of our respective experiences. Unfortunately for the cat, that content could not be the content of feline judgement.

So it would be if the sameness here must emerge in a sameness of representational content. And this must be so if the information about my surroundings which my experience makes available to me must be made available through things being represented as so. For what it would take nonconceptual content (in Peacocke's sense) to represent as so is what I might recognise to be so of my surroundings, given what I see of them – that the blood from the T-bone has made just that pattern in the rug, say. But we are not so shackled. I and the cat both see how things are in and about that bush. We thus see, and experience, the same thing. We may both, say, be responsive – each in his own way – to those avian movements in the branches. I can recognise things being that way as their being thus and so, for various values of ‘thus and so’. The cat cannot do all I thus can. But sameness of experience here need not lie in sameness of how anything is represented in our shared awareness of the scene before us looking as it does. Without the SH model, it can just consist in the awareness afforded both of us of the scene being, visibly, as it is.

Which, on reflection, is no doubt a good thing. The scene before the cat and me is, in being as it is, literally innumerable ways there are for a scene to be – one for each way of reaching to a range of cases which reaches this one. I cannot grasp all the ways (condemned as I am to the human condition). It is hard to see how my experience could represent things as all these ways. So there are our friends the Martians, graspers of many ways beyond our grasp, failing to grasp others within ours. With them in view, it appears that all content of all our judgements is nonconceptual in the sense suggested by Peacocke's remark above.

Still, there is an underlying picture here which needs to be addressed. It is clearer in Evans than in Peacocke. Their shared idea was: in visual experience, things appear to me a certain way. In Evans things so appearing makes it recognisable to me what way, or ways, my experience represents things to be: those ways it is as though things are in things appearing as they thus do. Now a thought might be: Why not cut out the middle men? If things appear to me a certain way, let my experience represent things to me as that way. Things appear as though blah; my experience represents them as blah. Here original tension emerges in a new way. Things appearing to me a certain way can just be things appearing to me as they then do – a denizen of the nonconceptual. So we have here something nonconceptual in the Frege-inspired sense I have given that term. The way my experience represents things is (something like) as they appear. If their so appearing is their appearing as they do, so something nonconceptual, then, it seems, so to represent them is to represent them as a certain way, where things being that way also belongs to the nonconceptual.

But the ‘certain way’ in ‘represent as a certain way’ cannot be read in the same way as that in ‘things appear a certain way’, where this is to be read just as things appearing as they do. It must be read as such-and-such way there is for things to be. Otherwise we have no representation at all. If anything is represented as so here – if there is anything truth-evaluable – then things need not be just as they in fact are for things to be as thus represented; nor for them to be so represented. The particular case is represented as what it would be in doing something (determinate) less than just being all it is; so in what might be done in a range of cases. The idea is things appearing as they do, just in being what it is – a certain particular case of things appearing – is to determine how this representation is to reach; just what it would be for things being as they are to be all they need to be to be as represented. But particular cases, have no reach, determine as such no reach for anything conceptual to have. So the idea makes no sense.

What could give the impression that merely things appearing as they do, in itself, while remaining nonconceptual, could, for all that, determine something it would be for things to be as they were according to it – a bit of the conceptual? Evans suggests one thing. An ordinary photograph registers (digitally, on film, in a print) the scene before the lens. It thus contains information about that scene in our now familiar first way. There is that photo of Pia on the boardwalk. It shows the way she was back then. Her hair was bobbed then, her chin line firmer. There was that sparkle in her eyes, now lost. There is much the photo does not show: Sid's sideburns and bellbottoms (he took the photo, was not in it); his then-addiction to chicken nuggets; the lovers under the boardwalk. Nor whether Pia was suffering indigestion, or her obscured arm was then freckled. So things would have been as shown in the photo while the world at large was – things were – any of various ways. In a range of cases things would have been as shown – perhaps the range some bit of the conceptual might reach.

So the photo shows Pia to look a way she could have looked while various things were going on. Someone might take that to mean that it represents her to be that way – that there is a way things are according to it. If photos can do this, it does not seem to require conceptual capacities so to represent things; which may want to make us label representation so achieved as ‘nonconceptual’ (though the achievement would remain presenting the particular case as falling under some given generality).

One tempted by such thoughts should attend to what answers questions as to when things being as they were would be their being as shown. Would Pia's hair have been bobbed then if things were as shown? One might consult the photo to see how it presents her. But is what it shows hair being bobbed? A photo cannot answer that. What lies in, or behind, what a photo thus cannot do?

If the photo represents things as so, the most it could do is represent it to be so that Pia then was (inter alia looked) thus – as shown. If being thus counts as having one's hair bobbed, then her being as she was according to the photo is, in fact, her having her hair then bobbed. When it comes to which concepts under which to bring Pia's being as thus represented, just how being as shown reaches beyond the particular case, and to what, what the photo shows, its photographic image, so far as that goes, leaves us entirely on our own. Someone might think of this as just giving further content to the idea of nonconceptual content. The thought would be: perceptual experience is just like that. It, too, represents as scene as thus – as it thus appears. But we need to look more closely at whether this is really representing at all.

We began with the idea that a photo contains information as the nonconceptual does. This has been worked into the idea that it, so experience, too, can do that while representing as so. It contains information retrievable in the first way, but in representing things to be such as for that to be retrievable information. What, then, decides what information the photo thus contains? The photo contains information as to how Pia looked back then, or at least that day on the boardwalk. Obvious first questions: Back when? What day? What answers these is when the photo was taken; not, at first blush, information the photo contains. Perhaps the information is written on the back (if a print is in question). Otherwise, there is an historical fact, to be sought as such facts are. E.g. ask Sid. By contrast, suppose I say, ‘Pia wore her hair bobbed then’, saying Pia's hair to have been bobbed at a certain time (or in a certain period). When was her hair bobbed according to me? That depends on how my words are to be understood. For all of my uttering them when I did, I may have said, or not said, her to have had bobbed hair at any of an indefinite variety of periods, or moments.

The crucial point has been made. In a court of law we would speak of the photo meaning (factively) that, say, Pia wore her hair bobbed when she was 17 (so, further, that she was not then in the cult). Which is just what it might do, depending on the facts: when it was taken, how constant Pia was in her coiffure in that period, whether she might have been wearing a wig. A caricature of Pia with bobbed hair, or a photo of her (or a body-double) with bobbed hair, as one frame in a photo-BD titled, ‘Pia at 17’ might mean something as to her coiffure at 17, or her coiffure at 35 (if, say, her flowing tresses then would have made the caricaturist depict hair as bobbed). But it represents her hair as bobbed at a time decided by how it is to be understood. For the BD, produced when she was 35, using a body-double, the title is a clue to that. These are all things a photo as such could not do. There is nothing wrong with the idea that that caricature means that Pia's hair was bobbed at 35, but represents her as with bobbed hair at 17, or that it represents her as with bobbed hair at 17, but does not represent her to have had such hair. All things beyond the reach of a photo. Moreover, that frame of Pia in the BD might represent her as with bobbed hair even if she is wearing a wig. If I said, ‘Pia had bobbed hair’, the fact that she might have worn a wig in those days takes nothing away from the fact that I represented her as having bobbed hair; it merely means that what I said might be wrong. Again, if the photo happens to be of a body double, then all it shows is the body-double having bobbed hair, and all it means (pending further facts is that the body-double's hair was bobbed. Such are some of the differences between representing as so and factive meaning.

The differences are manifest throughout. In the photo, Pia's skin is smooth and pale. Her eyes seem preternaturally wide. Does the photo show her as having such skin, or eyes? It depends on how the photography worked. Perhaps the film (or programme) is not true to skin colour, the lens not true to eye dimensions. The camera produced a certain image; now the question is what, given the apparatus, such an image means; what we can conclude from it, what information is retrievable. If the film is not true to skin colour, then it is not as though the photo represents Pia to have skin of some colour other than she did. Rather, to that extent, it contains no information as to her true skin colour. It cannot represent falsely; by the same token it cannot represent at all.

Factive meaning is a relation borne, sometimes by the conceptual, sometimes not, to bits of the conceptual. So there is, as a rule, a range of cases in which things would be as meant. But what factively means cannot, in doing that, reach in any particular way to any particular range, as is intrinsic to what belongs to the conceptual. Photos mean, but cannot represent (though, like anything, they can be used to do so, as in the BD). They are no model for how anything could represent nonconceptually.

5. Filling Out Space

The SH model cannot allow perception to bring the world to bear on what to think as perception would do. Representation cannot allow it to. So the SH model of perception is wrong. Representing in or by perceptual experience – notably things as so – requires the SH model. So there is no room for representation in or by perceptual experience. The next question is: if the SH model fails, what is the right way to conceive perception? But temptations to think that perceptual experience represents things run so deep, and touch on things so fundamental, that it will help in addressing that question first to consider, from some other angles, why this is something perceptual experience could not do.

Peacocke's idea of how visual experience represents things is, seemingly, inspired by that photographic (mis)conception of how something belonging to the nonconceptual can be conjured into a bit of the conceptual; perhaps by a conception of how a photo can be digitalised. Accordingly, he postulates a kind of content of such experience which he calls scenario content. This is content of a visual experience, identified by a scenario. A scenario is a 3-dimensional space with labelled origin and axes, and with each point in the space assigned some set of properties drawn from a stock out of which scenarios are generated. A positioned scenario is assigned a viewer at a time. For example, if the origin is right between the eyes, and the axes right-left and up-down, then to position the scenario for Sid now is to take the origin to be right between Sid's eyes, where he is now, the right-left axis to run from Sid's right to his left, facing as he is now, etc. Sid's present visual experience has the content thus identified just in case things are as they are according to the experience only if each point in the space around Sid has the properties assigned that point in the scenario. One might, in that case, call the scenario true, or true relative to that positioning. ‘Only if’ because scenario content is only one sort that Peacocke thinks visual experience might have. One might say: on this condition, Sid's experience is true so far. A wrinkle: Actually, Peacocke tells us, the content of Sid's experience is identifed by a family of positioned scenarios. His experience is true so far just in case one of them is true. The family is to reflect Sid's inability, e.g. to distinguish fine shades of colour. I will ignore these complications except where they demand attention.Footnote ²⁵

So if a given positioned scenario fits Sid's current experience, then that experience is true so far if that positioned scenario is true: for that scenario to fit is for there to be a certain way things are according to the experience; for it to be true (so positioned) is for things to be that way (though this may not be the whole way things are according to the experience). Indeed, this is what it is for a scenario to fit. If Sid's experience represents things as they are according to the scenario, it fits. There need not be a unique scenario which thus fits. Perhaps there is a unique one generated from some given stock of properties. But there may be others generated from other stocks. If a scenario which fits assigns to a given point being blue, and part of a solid surface, that is one way of capturing how things are according to Sid's experience. But there may be others. (This is part of the inspiration for nonconceptual content.) The question now is: Which scenarios would fit Sid's experience, given that it was as it was? About this Peacocke says,

Really? Let us see. First, though, for some ground rules. Little in them should be controversial here.

First, how much should a visual experience should represent as so? Suppose can just see Sid's inebriated look, the red meat on the carpet, the blood seeping into the white fibres. Then, perhaps, that visual experience should represent it as so that Sid looks, or is, inebriated, etc. But suppose she only learns he is inebriated later, by seeing the bottle of Old Codswallop he has been at. She may deduce, or infer, from that, correctly, that he is inebriated. But this is not part of what she is, or is made, aware of in, or by, seeing Sid staring sheepishly at the red meat on the rug. So it is more than that experience should represent as so. A visual experience should not represent as so what one would need to deduce from other things one had taken in. It would thus proffer more information than vision, or mere seeing, then makes available. (With rules to come, this should mean that Pia's present visual experience should not represent more as so than she could, or might, see to be so if her experience is in fact one of seeing a scene before her.)

Plausibly, for Peacocke or Evans, some form of converse point should hold. If Pia just sees Sid going at the Old Codswallop, and can thereby see that he is going at it, then her experience should represent this as so. For, though seeing that need not be based on visual input, here it is. Here there is nothing else from which Pia must, or could, infer that Sid is so engaged. Inference is not in the cards here. Pia can just recognise of what she sees – the scene being as it is – that this is Sid going at the Old Codswallop. Peacocke tells us (and I think Evans agrees) that experience representing things as so is the way in which it allows us to learn things from it as to how things are. This, we know, is false. But if it were true, then, since Pia's current visual experience allows her to learn, just from it, that Sid is so engaged, it would have to represent this as so. Given its ad hominem character, I will treat this as plausible, but not mandatory. But if experience representing things as so is not the only way for it to make the world bear on what to think, what is the point of its doing so at all? Such would just be making the same information available in two ways at once – as where Pia, really, really, not wanting Sid to forget their anniversary, sews the place with both wedding pictures and notes.

Second rule. It should be recognisable to the subject what way his visual experience represents things to be. Something in what he is, or can be, thus visually aware of should make this recognisable to him. I am not here disputing Evans' claim that perceptual experiences cannot themselves be objects of perceptual awareness. Nor am I supposing that perceptual experiences bear their content in anything like the way a sentence does. But I recall something Evans himself relies on: enjoying a perceptual experience is a way of experiencing being in whatever state enjoying that experience is – being in one of Evans' internal states, if such there be. What I am insisting on is: experiencing what one thus does – something one thus experiences – makes it recognisable to one which way one's experience represents things to be. What one experiences can just be things looking to him as they then do. Recognition need come to no more than this: confronted with the way things relevantly are (or prove to be), one is able to say (see) that this is not, or, again, is, things being as they were according to that experience: the experience was wrong, or right. And one can recognise, sufficiently often, of sufficiently detailed concrete examples, what would be, or not, things being as represented. For only recognisable representing (in the present attenuated sense) could make the world bear, for the subject, on what to think and do. Which was to be the point of representing in experience.

Third, it cannot follow merely from the fact that the scene before me is not a certain way that my experience did not represent it as that way; nor, conversely, merely from the fact that my experience did represent things as such-and-such way that they are that way. For experience, as elsewhere, truth must not be a requirement on representing at all. For suppose that my experience could not so much as say that there was a sloth in that tree unless there was one. Then it could not recognisably so claim where it was doubtful that a sloth was present, or where this was something one still needed to learn. So experience could represent only what there were other ways of ascertaining, or being certain of. Experience would thus reduce to a fillip on what fares well without it.

Fourth, if experience may represent falsely, it cannot lie blatantly. Suppose I plainly see the snake moving in the grass. Then my experience cannot represent it as so that nothing is moving in the grass. Which means: if my experience represents it as so that the snake is moving, I cannot tell just by looking (at the snake) that it is not. My visual experience cannot represent it as so that the snake is moving if things look to me (so far as I can tell how things look) just like a snake lying still. I omit a general formula, but suppose we can recognise instances.

6. Commitment

The curious idea that perception represents overlooks many obstacles. This section sets out one; the next another. The first obstacle is pervasive. It arises wherever there are two ways for things to look the way they do. Suppose I face an effigy of São Mamede. It looks blue. Suppose there are three ways for it to look just the way it does: it may be made of blue plaster; it may be painted blue; it may be neutral-coloured, coloured blue by lasers. (The example, inspired by laser-coloured effigies at Amiens, is not far-fetched.) Perhaps things would not look just as they do all three ways: you could tell if the effigy were made of blue plaster, rather than being painted. In that case, representing is de trop: its looking as it does means that it is painted. But suppose things might look just the same one way or the other. By our rules, what I experience visually must make it recognisable to me how things are represented. But how could things being visually as they are for me make it recognisable to me that visual experience represented things as one of these ways (blue plaster there), rather than any of the others?

The effigy looks just like a blue-plaster effigy – as one would or might. It must be a blue plaster effigy to be as it thus looks. It looks just like a blue-painted effigy. It must be blue painted to look as it thus looks. It looks like a blue-laser-illuminated effigy. It must be that to look as it thus looks. But none of this so much as speaks to the question how things must be to be the way they look (‘things’ here bearing its catholic reading, on which one cannot ask ‘Which ones?’). Yet, it seems, visual experience provides nothing beyond things looking as they do to tip me off as to how it represents things to be. So it seems that such experience could not represent things as any of these ways. Of course, I may be in a situation in which there could not be lasers, or, perhaps, blue plaster. In that case, things looking as they do might mean that there is blue paint there – but then, no representing needed.

A natural reaction here would be some form of minimalism. Let us retreat to it by steps. First, perhaps my visual experience just represents that region as occupied by something blue, no commitment to how blue. But if the effigy is neutral coloured, the blue thanks to lasers, there is an understanding of something being blue on which it does count as something blue, and a contrasting one on which it does not; on which something blue only in the laser light is not something blue. Similarly for paint. So on what understanding of being blue does my experience represent that region as occupied by something blue? The above form of argument can be repeated here.

Second, perhaps my experience merely represents it as so that things look blue in that region of space. But this cannot help. What makes it recognisable to me how my experience represents things (when to say it was false) is things looking as they do to me. There are two ways for that to happen: they may so look; or it may just be me. Now the above argument repeats itself again.

Third, perhaps my experience merely represents it as so that things look blue now to me. There are several problems with this idea. But one will do for the moment. I am already authoritative as to how things look to me. (Or if I have lost track of that, it is obscure how representation can help me.) Of what is representation supposed to be informing me here? Or is it, again, just a wheel idling. (Aunt Ida as I stand before the donkey: ‘That's a donkey.’) Perhaps it informs me that that (things looking as they do) is something looking blue. But then how am I to understand it? (It would be nice if my visual experience could represent it as so that that colour is heliotrope, that one burnt umber, and so on, I being bad at recognising those colours. It would then be a sort of cognitive prosthetic. But, though, taking that ‘that’ to refer to a region of my surroundings, it may be so that that is burnt umber, it is hard to see how experience could commit itself to any such thing. Recall our first rule: visual experience should not represent any more as so than is retrievable from what is visually experienced. If I do not know burnt umber when I see it, then that that is burnt umber is not information so retrievable.

So, it seems, there is just nothing visual experience could legitimately represent as so.

7. Status

Using the strategy of the last section, I now develop a second obstacle. I begin, it will seem to some, off target. I then hone my aim. Here the problem is squaring experience's representing with our first rule (and its converse). If I oversimplify epistemology, the main point will be seen, I hope, to withstand the complications. I draw it from Thompson Clarke (Reference Clarke, Black and Alston1965). It would not touch a view that visual experience represents, not the scene before one as such-and-such ways, but merely (in suitable cases) itself as revealing – as an experience of seeing – how things are. That idea was disposed of in the last section.

To begin, then, a fresh baguette is on the kitchen table, on a breadboard, facing Sid (seated). What does Sid see? One answer: ‘The baguette.’ If so, then perhaps, by 1's converse, his experience should represent it as so that there is a baguette before him. So, too, for any ringer for a baguette-seeing experience – say, one of seeing a play-doh ‘baguette’ – what visual experience would not reveal as not one of seeing a baguette. Suppose, though, that the right answer is, ‘The front surface of a baguette.’ Then, by the first rule, Sid's experience should not represent it as so that there is a baguette before him. That would be something for him to infer from the information perception made available. Such is the shape of seeing's bearing on what visual experience might represent as so.

Here Clarke's point: no answer is the right one as such to ‘the’ question what Sid (actually) saw. Rather, different answers would be the right ones on different occasions for saying what Sid saw. What counted, on some such occasions, as what he saw would not do so on others. And there is no further occasion-independent fact as to what he ‘really’ saw. Similarly, an undyed cotton shirt, coloured blue by lasers, counts as then blue on some understandings of what its being blue would be, as not blue on others. There is no further fact as to what its colour ‘really’ is. Besides those occasional ways of drawing a blue-non-blue distinction, there is no other.

The Clarke point, as I will call it, is that there is no one right answer to the question what Sid saw. So there is no one right answer to the question what he would need to infer. So there can be no one right answer to the question how, by our first rule, his experience is to represent things. Neither the samples above, nor any other answer could be right as such. Which means that there can be no right answer at all. If Sid's experience is to represent, for him recognisably, then how it represented in a given case cannot vary from occasion to occasion of the asking, according to what Sid would then count as needing to infer.

One could argue for Clarke's point by cases. If the little imps have been attaching front surfaces of baguettes to play-doh and leaving them in hopes of their muzzy elders preparing play-doh tartines; if such impishness may have struck, then, while Sid may say he saw a baguette on the breakfast table, we might count him as really having seen only a front surface. Once the children have all been packed off to military school, we can go back to talking about seeing baguettes. What Sid saw in not decided merely by the presence of the baguette, nor by his alertness being as it was to the occupation of that region; nor by whether the children have in fact been packed off. The circumstances of discussing his situation matter.

But there is a better way to see the point. If perception's job is to make the world bear on one's thought, then to see something is to enjoy a certain epistemic status. That thought, worked out, imposes occasion-sensitivity. I start with perhaps slight oversimpliction. An experience of seeing a baguette before one is one which allows the world to bear on his thought as a baguette's presence before him bears on what is so. Which means: it makes available, to its subject, the information that there is a baguette before him: given adequate, and functioning, recognition capacities, he can recognise the scene before him as one in which there is a baguette before him; so he can judge that there is one there on grounds that he sees it. Seeing can be, for him, proof. There is no gap between seeing a baguette and there being one through which absence might slip. But if one only sees the front surface of a baguette, then no more is made available to him than that there is a front surface: judgements as to baguettes must be conclusions drawn.

Deciding what someone saw is thus deciding how the world has been brought to bear on his thought; what about it has been made available. Cases like the above bring out the need for such decisions to take account of the uses to which such information is to be put. Telling whether there is a baguette or play-doh on the table may be one such use. For Sid to be thus enabled would be for him, then, to be able to tell baguettes from play-doh at sight. Where clever imps are an issue, he cannot. Ordinary recognition capacities – say, knowing a pig at sight – are inherently dependent on hospitable environments for being that. If there is a tapir which looks just like a pig, or a pot-bellied pig which looks very unlike one, and if the gentlemen farmers of Oxfordshire are stocking their farms with these, then, though, usually, I do know my pigs when I see them, I cannot tell a pig by sight in Oxfordshire. It is not the actual one-upmanships of Oxfordshire gentry which cost me the capacity. False rumours in Chelsea may cost me it, on some occasions for the reckoning, if Chelsea somehow relies on me. Whether I have lost my capacity on arrival in Kingston Bagpuize all depends on who is asking, and when and why. Knowlege would not be in the picture otherwise. All the rest follows from seeing's role in making one knowledgeable.

Such is a simple thought. Perception allows information to bear on what one is to think – given his ability to extract it from a scene by his capacities to identify what counts as being what. Sid sees that goldfinch with its distinctive goldfinch head. He cannot tell it is a goldfinch: he does not know them by their heads. Seeing may well (though need not) make the information available (on one good understanding of available). Perhaps Sid could describe the head so that an expert knew at once it was a goldfinch. Still, that there is a goldfinch on the branch is not something that may yet bear, for Sid, on what he is to think. In the bread case, the problem is not knowing what distinguishes a baguette; nor with what distinguishes one from Play-doh. It is that in impish surroundings the marks of a baguette are not on its facing crust. So it is not that perception makes available to Sid information he cannot use. It is rather that it does not make available to him information he perfectly well could use if he had it. It thus does not make available to him the information that a baguette is before him – not even as it did that there was a finch on that branch. So he cannot have seen the baguette. Now the question is: for purposes of saying what he saw, what count as the circumstances he was in – impish or not? The answer to that question can only be an occasion-sensitive matter.

Can this be right? If I have checked that it was, in fact, a baguette on the table, I might describe Sid (truly) as having seen a baguette, though all he could tell was that there was a facing surface. On the other hand, in impish surroundings, if Sid claimed to see a baguette, we could confront him with the impish possibilities and force the admission that all he really saw was the facing surface. All of which is just what one would expect if it is an occasion-sensitive matter what circumstances Sid counted as being in. When I establish that it was a baguette, I establish that Sid was not in (relevantly) impish circumstances. I now speak of what he saw accordingly – as it is to be spoken of on an occasion where his circumstances so count. This no more shows that what he ‘really’ saw was the baguette than finding an occasion on which one would not say a shirt was blue shows that it ‘really’ is not blue. It does not, for one thing, because it no more shows this than confronting him where he counts as in impish circumstances shows that what he ‘really’ saw was a facing crust. What one sees, so what one's experience could, or should, represent as so, depends on the circumstances one is in. What these are depends on when, and why, you are asking. Which destroys the idea that there could be such a thing as a way things are according to an experience of seeing.

Which may tempt a representationist to minimalism. The rough idea would be: what an experience of seeing represents as so is just the least one would ever count as seeing (if one is). If there is a least Sid would ever count as having seen viewing the baguette, this would have to be a facing surface, or region of space. Beyond that, the environment drops out and it is seeing no longer. But the Clarke point applies to surfaces and regions. Suppose Sid, lost in reverie, is but dimly aware of the baguette. On the crust is a dark spot which some say resembles São Mamede. Here cracks, there what might be an impish thumbprint. Such details are lost on Sid. Did he see the crust? Or only part, or parts, of it? If the last, then which one(s)? Just which surfaces before Sid's eyes are ones he could judge to be present, or thus and so, which regions thus and so, on grounds of seeing their presence, or being that way – that is, non-inferentially?

Suppose we (mentally) place a grid over the facing crust, one square, e.g., taken up with the São-Mamedeish spot, and ask, for each square, whether Sid saw what is in it. Could Sid really judge, of square R6, that that was occupied by crust – on grounds of seeing it so occupied? Suppose imps may have been cutting bits out of crusts, hoping we will suspect mice. Does what Sid sees allow him to say whether R6 is occupied (or even base a judgement on how he has seen R6 to be)? One would, at least, sometimes, suppose not. But this device, if it can be worked at all, can be worked for each region of the grid (not necessarily simultaneously). (Things get worse if, to oblige Peacocke, we try for answers for each point on the crust.) If we add up these results, we would seem to get the result that Sid sees none of the facing surface (or at least not much, or much of it not), which would be not to see it at all.

But suppose we now ask whether Sid, as he then was, was enabled, by sight, to base a judgement that there is a front surface before him on the presence of that surface – whether he might be able to judge such a thing non-inferentially. For each bit of grid, or most, we can build a story within which it would be wrong to say that Sid could judge non-inferentially that there was crust in that region, simply, by his seeing it, on grounds of there being crust there. But this does not seem to preclude a positive answer to the question. Seeing what he did may make Sid authoritative as to the presence of the crust, while being liable, for any, or even many, regions of the grid, to leave him not authoritative as to whether there was crust there. It may be liable to grant such authority over a region even where, if there were no crust in that region, then there would be no facing crust present (but only, at best, a part of one). Which forces some form of occasion-sensitivity: Sid may count as authoritative as to whether there is a crust; basing judgement on the presence of a crust, even where we recognise that, on certain occasions for asking, he would not count as authoritative as to the presence of a crust in region R, and if there is no crust in region R, then there is no facing crust for him to see. He counts as authoritative as to the crust just where his circumstances do not count as ones in which he is not authoritative as to region R (which need not be for them to count as ones in which he is so authoritative). Now we need only remember: that kind of occasion-sensitivity is incompatible with squaring experience's representing with the demands of rule 1.

Minimalism is more familiarly worked in a slightly different area. Sid may wrongly, but understandably, take himself to see a baguette where there is only a facing crust. All he really saw was the crust. But it seemed that there was a baguette: so far as he could tell by looking there was. Similarly (it seems), Sid may wrongly, but understandably, take himself to see a facing crust where it is just a facing image, or just all in his head; but where there is nothing he experiences visually to tip him off. (It is none too easy to say how it might be all in his head. But let that pass.) Where there is only the facing crust, it seems that he is seeing something he would, or might, also have been seeing where there a baguette. By parallel, it may seem that where there is no facing crust, he experiences visually (seeing now having dropped out) something which he would also experience visually if there were a crust, or, still better, a baguette. This something-visually-experienced, whatever it is, would be the minimal ingredient in an experience of seeing, present independent of the occasion for the asking. The rest can all then be seen as mere seeing-by-courtesy.

The manœuvre can be made lose its charm. Sid sees the baguette. What is it that he would see whether there was a baguette or not? One answer is: the facing crust. But does Sid see that whole crust, or only part of it? That is an occasion-sensitive matter. What was it like visually to see what Sid did? One answer to that would be by holding up the crust. Things – what he saw – looked like that. Here, one needs to fix an understanding of looking like. There are understandings on which the crust, when I hold it up, no matter how carefully, does not look as it did when Sid saw it. (Light and shadows, for one thing.) But anyway, if we aim to find a minimal feature in visual experience by looking inward, to, say, the features of some internal state, something belonging to Sid's consciousness, as per the SH model, then holding up the crust does not even address the question we mean to ask. What we what to know is how things were visually for Sid. How, then, was it visually for Sid if there was a São-Mamedeish spot on the crust he saw? If pointing to the spot on the crust is not an answer, then there is no answer to this until something fixes what it would be for things to be one way or the other in this respect. That would be work of an occasion for the asking. Without that – unless, perhaps, the spot was, for Sid, particularly striking – there is no such thing as ‘how that region looked to Sid’. In general, there is no such thing as the way things looked to someone in his seeing what he did, unless that way is simply: the way they did. So this way of moving inward moves nowhere. It is crucial, if experience is to represent, that it lead somewhere. For, to repeat a now-familiar theme, only determinate objects of visual awareness other than what is seen could possibly make room for experience to represent.

8. Gazing Out

It is not for a philosopher to deny that retinal images unleash chains of happenings. Personally, I am inclined to believe this. Nor is it for a philosopher to deny that such chains terminate in states of some characteristic sort. I am officially neutral. Nor is it for a philosopher to deny that without such happenings and states we would see nothing. Some such things no doubt enable seeing. This settles little as to what it is that they enable. It does not settle what someone sees on an occasion, or when someone would count as seeing such-and-such. Nor does it help much to insist that the things he sees must be ones whose images on the retinas unleash things. Nor can we suppose that those terminal states in the enabling story just envisioned, in having the features they do, might ipso facto settle something which was the way things (then) looked to their possessor. Nor, correlatively, should we suppose that they settle something which, with the presence of given further factors, not themselves seeing, or even visual awareness of anything – pre-retinal aetiology, say – might add up to seeing whatever their possessor happens to see on the occasion – as if seeing were a hybrid of appearings and a friendly world.

How, then, to conceive visual experience? It has taken too long to get this far for me to say very much about this here. But what has been said so far provokes a few suggestions. First, some visual experience is seeing, some is not. Seeing is what is enabled when all goes well – as all is liable not to. If we do not conceive seeing as a hybrid, with an underlying ‘way things look to the perceiver’, present whether things are going well or not, we might then treat these cases separately. Which, I suggest, would be helpful.

I begin, then, with seeing. One striking difference between seeing (perceiving in general) and testimony is that seeing is, and remains for its duration, au courant. If Sid is told, reliably, that his ribeye has just gone on the grill, he may then know how things were then. So long as he has the testifier's word for it, he, perhaps, continues to know how things were then. If, meanwhile, the ribeye has been carbonised into inedibility, it takes a new piece of testimony to make him au fait with that. If, by contrast, Sid sees his ribeye going on the grill, he is thereby afforded, for so long as the experience lasts, constantly au courant awareness of the ribeye's career. He can suffer agonies as, helpless, he watches it carbonise, or, a bit less feckless, flip it himself. Seeing is dynamic. (So, perhaps, are enablings, though, again, my philosophic business is not to claim, or hypothesise, so.)

Seeing things is thus following them. Sid, before the baguette, shifts his attention, his eyes, his head; fidgets, swivels, rolls the chair around (it is one of those). Whatever we say as to his seeing the baguette, we do not normally conceive of seeing such that the right answer to that question changes from moment to moment – unless there is some special reason for it to change, as, when Sid swivels all the way around, the baguette goes, momentarily, out of sight. So seeing the baguette is something one qualifies, or not, as doing over an interval. It is relevant intervals – whatever counts, on an occasion, as Sid's experience of seeing, or failing to see, the baguette – which are evaluable as ones in which he saw it, or, aain, did not. More reason not to suppose such a thing as ‘the’ way things looked to Sid in seeing the baguette, or that the momentary features of some internal state could fix what this was. Here, too, we find reason for conceiving seeing sometimes as affording awareness, if also sometimes as actually conferring it (Sid may see (still) the baguette at a moment at which his attention is fixed on that dark spot) – but, in either case, as a matter of what Sid holds onto throughout the interval.

A second central feature of seeing (and perceiving generally) is that it is occasion for exercise of our capacities for recognising the reach of various bits of the conceptual to various cases of the nonconceptual. For being carbonised to reach to the nonconceptual as it does may be for it to reach to Sid's ribeye's being as it is: Sid's ribeye's so being is a ribeye being carbonised. The very thought that that is a ribeye being carbonised is not one which Sid, or anyone, could so much as entertain without acquaintance with that – the particular case (any more than I can now think, of a certain man in Ulan Bator, that he is now drinking tea). Seeing the ribeye is an exemplary way of making that thought available. Such is a most central role of sight – tied up, too, with the idea, discussed above, that seeing has really not made the world bear on what to think as it ought if it falls short of providing acquaintance with the particular case – the ribeye being as it is.

More generally, seeing makes thoughts, and, more generally, attitudes available to a subject which are otherwise liable, sometimes bound, not to be. No doubt these extend beyond those involved in seeing the reach of the conceptual to the nonconceptual. So we now have two measures for what it is that someone saw: what one sees must stay in step with one's epistemic status, as per the last section. It must also stay in step with the attitudes one might count as holding. Seeing the man in Ulan Bator allows me (ceteris paribus) to think of him that he is drinking tea. Conversely, the right thing to say as to whether I can so think may bear on what to say as to whether I saw him. These two constraints may allow one to look at a subject, and what is before him, and, without looking inside him, say, on occasion, what it is he saw.

Now for visual experiences which are not seeing. I confine attention to cases where enabling processes and states have gone wrong, as where, in special conditions of illumination, convex and concave, e.g., on that house across the road, may reverse. I begin with a problem. We can think of seeing as affording, as well as as conferring, awareness – awareness one might enjoy, where one might also have failed to; awareness of what there is anyway to be aware of. We can do that precisely because seeing is a relation to an environment, whose denizens are for one to encounter, broadly speaking, in experience. It is just this – that awareness is of what one might be unaware of – which calls for enabling, such as might be provided for seeing by chains of happenings unleashed by the retinas, and so on.

The task of those chains of happenings unleashed by retinas was to be to enable visual awareness of one's surroundings. The SH model assigns them another role: to issue in states affording awareness of something else: how things are visually for – how they look, or seem, to – the subject, where this is independent of how things (pleonastically, in the environment) are visually, or look, or seem. This goes inevitably with the idea of such a state as with other features (physiological, perhaps) which impose a unique answer to the question how things are visually for one in it. If such a state thus affords awareness, as distinct from conferring it, then this is awareness one might enjoy or not (as, e.g. Evans, insists). So it is something which calls for enabling, just like seeing itself. And, to borrow a term from Frege, the game can begin anew.Footnote ²⁷ (Nor could the enabling here be self-enabling, accomplished by those very processes and state which provide such further things to be aware of: such would not be for us to be afforded awareness of them.) It is not quite incoherent to suppose that an enabling internal state produces something its subject must, but may, be enabled to be aware of. But it is incoherent to suppose it an intrinsic feature of enabling stories that they involve states which do produce such things. It would at least eliminate promissory notes if the story of what enabled seeing did not posit any such thing.

Suppose that what enables seeing also, on occasion, provides something else of which to be aware visually – say, things looking as they then do to someone. Suppose that this something else belongs to the contents of that person's consciousness, in Frege's sense. Then Frege's lesson applies: such contents of consciousness could not be objects of a judgement; there is no sense in the idea that they, just in being as they are, might be thus and so – might thus relate to some given bit of the conceptual.Footnote ²⁸ Which does in the idea that one might be afforded awareness of such things – as if that is awareness one might enjoy or miss out on.

The idea of seeing as dynamic suggests a way of reading Frege's lesson. On that idea, a stream of impressions on, so chains unleashed by, retinas, over an interval, goes with a certain responsiveness to the scene before the viewer's eyes, stable over that interval – stable opportunities for attitudes, or affordings of them. What one sees is then some sort of resultant of such responsiveness and what there is for it to be responsiveness to. If, say, what enables seeing sometimes provides further things to experience visually, which are not things of which to be afforded awareness in the present sense, then that role of the scene in the above co-operative enterprise drops out. Nothing performs it. What may remain is a subject's responsiveness (through intervals) to things being visually as they are for him; where here awareness may be conferred, but not (merely) afforded. Responsiveness to this could not take the form of judging that (in being as it is) it is thus and so. It could still take the form of seeing things being visually as they are for their being thus and so (auffassen, perhaps anerkennen; not erkennen). In any event, responsiveness could now take over the role of making such a circumstance one which, in being as it is, is a subject experiencing things being for him visually thus and so – e.g. as though he were seeing a cantilevered upper floor.

But suppose we think of what enabling-gone-wrong provides, not as contents of consciousness, but rather as determined by, say, features of some neural configuration, or, otherwise, facts as to how the visual system reverses convex and concave when it does. Now we can think of enabling as providing something of which to be afforded awareness. For now I am (so far) aware visually of what I am where, say, concave and convex reverse, in and by being a way there is for one to be. By the same token, now enabling is called for for me to be aware of what I thus am. Now, too, there is room for enabling to go awry, and in that way among others room for things thus to look to me one way or another in being afforded awareness of what I thus am. Again, the game can begin anew. But something must allow the chain of enablings to stop. The previous idea shows how it can.

All this is at best the bare beginnings of understanding perceptual experience. The main aim here was to clear away a picture of such experience which positively blocks understanding. It is a picture widely at work even in those who disclaim it. It is (most often) engendered by a false impression of what it would be to ask after perceptual experience in a scientific manner, or while giving science its due. As Wittgenstein predicted, that impression leads only into darkness.