Show Summary Details

Page of

PRINTED FROM OXFORD HANDBOOKS ONLINE ( © Oxford University Press, 2018. All Rights Reserved. Under the terms of the licence agreement, an individual user may print out a PDF of a single chapter of a title in Oxford Handbooks Online for personal use (for details see Privacy Policy and Legal Notice).

Subscriber: null; date: 20 August 2018


Abstract and Keywords

Typically philosophers who have thought that there is a distinction to be drawn between direct and indirect perception have thought it also obvious that we cannot directly perceive the objects around us which we suppose ourselves to know about through perception even if we do not perceive mere representatives of them in the form of sense-data, impressions, or percepts. For it has been common to suppose that if we perceive solid objects, we do so through perceiving only proper parts of them, their surfaces. The first section of this article focuses on the grounds for supposing this to be so and one radical strategy for resisting it. The second section turns to the persisting reasons for supposing that we cannot perceptually be directly in contact with the world around us. Whether the argument is successful depends in part on what we take perceptual awareness of ordinary objects to involve.

Keywords: perception, sense-data, impressions, perceptual awareness, ordinary objects, context-sensitivity

1. Introduction

Different views about perception are commonly distinguished by the claims they make about the direct or immediate objects of perception. Direct realists affirm that we directly or immediately perceive physical objects and that these objects are mindindependent. Indirect realists accept that there are such mind-independent objects and that we refer to them and have knowledge of them, but suppose us only to have indirect or mediated contact with them. Phenomenalists have commonly insisted that the objects of sense must be immediately or directly apprehended, and consequently have questioned whether there are mind-independent objects for us to think about in the first place. What, though, is it to claim that sensory awareness of some item is indirect rather direct? And what is the significance of the divide?

Presumably many readers know that Warsaw is the capital of Poland. Few of those who do know this will be aware of quite how they know it. They may dimly remember learning geographical facts at school, without recalling exactly the occasion on which they learnt this. They may know how to go and check the fact if their knowledge is challenged, and they might even know how to justify themselves in terms of general evidence about the organization of Europe. Still, neither of these is quite being able to identify the actual means by which they came to know or persist in knowing the fact in question. On the other hand, all readers of this page can discover that the first letter on the page is a ‘P’. But in this case, it should be clear how you can come to know this fact: you can just see that that's the case.

It may well not be true of most of our knowledge of the world that we have any access to exactly how we know these things, but in the case of many trivial particular (p. 702) matters of fact we often do know that our current perception puts us in a position to know these things. Correspondingly, the fact that some object is now perceptually salient to you is often a reason for you to single it out and think about it, or try to discover further facts about it. While not all of our perceptions need be conscious, our conscious awareness of the world typically is perceptual, and in reflecting on this we come to be aware of the objects and facts we have perceptual access to.

Now one might think that this picture of how perception, awareness, and knowledge fit together presupposes that perception of the world around us is direct: that what perception gives us directly is the ordinary objects of perception about which we form so many opinions. The importance of the question about direct perception, then, may lie in the role that our conception of how we perceptually relate to things affects our conception of how we come to think about and know about many aspects of the world around us.

Typically philosophers who have thought that there is a distinction to be drawn between direct and indirect perception have thought it also obvious that we cannot directly perceive the objects around us which we suppose ourselves to know about through perception even if we do not perceive mere representatives of them in the form of sense-data, impressions, or percepts. For it has been common to suppose that if we perceive solid objects, we do so through perceiving only proper parts of them, their surfaces.

In the first section of the chapter I focus on the grounds for supposing this to be so and one radical strategy for resisting it. In the second section I turn to the persisting reasons for supposing that we cannot perceptually be directly in contact with the world around us. Whether the argument is successful depends in part on what we take perceptual awareness of ordinary objects to involve. There are at least two possible ways of holding on to the idea that we have direct awareness of the mind-independent world—one an intentional or representational approach to perception, the other a disjunctivist approach to perception—and it is not clear whether we do have a determinate conception of what direct awareness must involve that will decide between these two approaches, or so I suggest in the closing section.

2. Seeing One Thing in Virtue of Seeing Another

Frank Jackson identifies a number of different threads in defenders of the immediate–mediate or direct–indirect distinction:1 one invoking the presence or (p. 703) absence of inference;2 that what is immediately perceived is that which is entirely known through perception;3 or that what is immediately perceived is that which is entirely perceived at a time.4 Jackson replaces these with a clear account of the contrast, an account applicable directly to the perception of objects rather than facts, and one that does not presuppose that there are sense-data or other non-physical intermediaries in perception:

x is a mediate object of (visual) perception (for S at t) iff S sees x at t, and there is a y such that (xy and) S sees x in virtue of seeing y.An immediate object of perception is one that is not mediate; and we can define the relation of immediately perceiving thus: S immediately perceives x at t iff x is an immediate object of perception for S at t …(Jackson 1977: 19–20)

Jackson explains that the connective ‘in virtue of’ here is not to be treated as a paraphrase of a causal connective, as when one says ‘He is angry in virtue of the lack of service in this restaurant’. It is, rather, used in the sense of showing some analytic or definitional relation between the two facts introduced by antecedent and consequent. The appeal to definition here could be loosened, perhaps by taking on Thomas Baldwin's suggestion that ‘ “p in virtue of q” is true where the fact that q explains why p obtains’ (1990: 240). We need to restrict this, though, in line with Jackson, to those uses of ‘explains’ in non-causal contexts.5

Moore and Broad assume that visual perception of physical objects is at least mediated by visual perception of their surfaces. Further argument or reflection may then show that such perception is mediated by non-physical sense-data (or in Broad's terms ‘sensa’) as well. Jackson, in common with this tradition, supposes that our seeing the surfaces of objects mediates our seeing the objects themselves:

We commonly see things in virtue of seeing other things: I see the aircraft flying overhead in virtue of seeing its underside (and the aircraft is not identical with its underside); I see the table I am writing on in virtue of seeing its top; I first see England on the cross-channel ferry in virtue of seeing the white cliffs of Dover …(Jackson 1977: 19)

Jackson gives us a brief piece of reasoning to this conclusion in the next paragraph. We cannot define perception of a part of an object in terms of perception of the whole object, because one could have seen the part without seeing the object (had the part been part of another object, for example), and one could have seen the (p. 704) object without seeing this part of it. So seeing the object can be neither necessary nor sufficient for seeing the part, and hence cannot be that in virtue of which one sees the part. This, of course, does not establish Jackson's point: even if we grant that one does not see the table top in virtue of seeing the table, it does not follow that one sees the table in virtue of seeing the top; one might just see both the table and the top, and neither seeing need be in virtue of the other. So Jackson relies tacitly on the assumption that there must be some constitutive link between the two facts. Why suppose that?

Jackson presents the problem about immediate perception as parallel to certain other cases in which one fact holds in virtue of another. For example, that one object touches another through one part of it touching a part of the other; that one is located in a country in virtue of being located in a city in that country; that an object is coloured in virtue of some part being coloured. In each example we have one fact involving a relation holding in virtue of another fact containing the same relation. So we might try to answer the question we just raised by generalizing the problem: what does one say back to a sceptic in relation to one of these other domains when he or she denies that there is any constitutive link between the facts at issue? For example, what could one say to someone who accepted that someone, Fred, is located in the United States and also accepts that Fred is located in Carson City but denies that Fred is located in the United States in virtue of being located in Carson City?

One might suggest that it is part of our semantic competence in talk of the location of such concrete objects as people that the one locative fact has to hold in virtue of the other. Hence the sceptic must be revealing some kind of misunderstanding of what is said, or at least of how things can come to be the case. Is there any way we can highlight what has gone wrong on the sceptic's part? As in the case of seeing, it is not appropriate to assume that one is located in one place in virtue of being located in another, and then determine that one of these facts must obtain in virtue of the other. For someone sceptical about the idea that one occupies some locations in virtue of occupying others may just deny that facts about an individual's location in one place need hold in virtue of the individual's location in any other place.

There are some things for us to add to the story. There is not simply one way of occupying a region of space. In general, we have a conception of the ways in which objects such as tables or chairs, or human beings, can occupy particular regions of space. So, for example, a human being will generally displace other solid objects from a region of space occupied by him or her. We can, thereby, make sense of the minimum region of space that such an object occupies through excluding from that region any solid entity entirely distinct from it. It is this sense of location that most concerns us when we wish to fill a Mini with students, or determine the number of items a passenger can place in an aeroplane locker. In turn, our understanding of what it takes for an object to occupy a region of space in its entirety intersects (p. 705) with our understanding of topology and the ways in which sub-regions of a space can be entirely enclosed within that space; and our understanding of geography and politics, which allows certain regions of land thereby to be included as parts of other regions.

In terms of these further claims, we can have some sense of truths that obtain independently of ascertaining the truth of Fred's being in the United States which seem to suffice, given our normal understanding, for Fred so being located once we grant that the region in which he completely excludes other physical bodies is in a certain space within Carson City. The sceptic's lack of competence would then seem to be revealed in an ability to make the move from this set of uncontested truths to the claim that Fred is located in the United States. Or, alternatively, given that the sceptic does not deny that Fred is in the United States, the lack of competence may be revealed in a failure to grasp how one thing can be so in virtue of another.6

In Jackson's presentation of the account of seeing, immediately seeing an object is taken as primitive, and what it is for an object to look some way to one is explained in terms of either immediately seeing it and it being the way it looks, or immediately seeing some other thing which is in fact the way that the first object looks to be (and hence in virtue of seeing which one sees the first). The ambition would therefore seem to be to explain the other perceptual vocabulary of ‘looks’ in terms of ‘sees’, and hence to presuppose that we can explain the distinction between ‘immediately sees’ and ‘mediately sees’ before understanding what it is for something to look some way. If one adopts this strategy, there would seem to be no parallel story for seeing to the one sketched above for the case of location. Hence we would seem to have no purchase on showing what aspect of our concept of seeing the sceptic would be getting wrong.

If we gave up Jackson's explanatory ambition, we might instead attempt to illuminate immediate perception by taking seeing objects and objects looking some way to one as equally primitive notions through which we understand the other. The analogy with location would then be restored. The basic idea is that just as a material object occupies a region through excluding other concrete items from it, an object is immediately seen where it thereby looks some way to one, and occupies in part the visually experienced scene.

Note that this is not to claim that we can reduce or entirely explain what it is to see an object in terms of how it contributes to how things look to one when one is seeing. That would be a hopeless task. First, not all cases of things looking some way to one need be cases of perceiving, or even of perceiving objects: one can be hallucinating for example, or one may be seeing but not seeing any particular thing. Even where one does see some object and things look a certain way to one, still one (p. 706) can't read off from the fact that things looking that way depends on some object the conclusion that you see that object. Suppose one sees a continuous red wall. The wall itself may be composed of bricks, although the discontinuities between the bricks may not be visible. In this case how things look to one as one stares at the wall depends on how (parts) of each of the bricks are. Had the middle brick been green, for example, and still part of the wall, then the wall would not have been a uniform red expanse. At the moment it looks to you as if there is a red expanse before you, and that holds in virtue of each brick being red and that redness being visible. Do you thereby see any of the individual bricks? Not obviously so. After all, none of the individual bricks is segmented out for you in the visual array as a possible object of visual attention. Now take a case in which one suddenly sees through a slit a vivid flash of turquoise. Behind the slit someone has walked past wearing a turquoise scarf. So how that person was, in wearing such a scarf, and how that scarf was, being turquoise, were responsible for how things looked to one. Nonetheless, we can't simply determine from that whether one has seen the person, or indeed the scarf. Indeed, in this case it seems that what we are inclined to say very much turns on the context: we can imagine a case in which you know that the given individual is liable to wearing such bright colours and that no one else is, so that in the circumstance you would be so placed to determine by sight whether that person had walked past. Against such background assumptions, we can construe a context in which one can truly say one saw the person; as well as a context in which one is prepared only to say that one saw a flash of colour. So there is no simple rule for moving from facts about how things look to a subject to claims about what things they see. We shouldn't hope to explain what it is to see some object in terms of a more fundamental idea of what it is for things to look a certain way to a subject.

Yet that is not what we require given our current purposes. We cannot define what it is for a physical object to be located in a region of space through the notion of excluding other matter. But the connection between the two notions helps us understand how one's location in one region must be constituted by one's location in a sub-region. In parallel, by taking seeing and things looking to one a certain way both as equally basic we can hope to elucidate how the seeing of one object may be constitutive of the seeing of another and thereby the distinction between mediately and immediately seeing.

So consider the kind of case often appealed to in discussion of why one could only be seeing an object through seeing its surface. We compare two situations: in case 1 you see an orange on a table in its full glory; in case 2 you are similarly placed and the surface of the orange is directed towards you as in situation 1, but the rest of the orange has been eaten away. There is some thing which you see in both cases which is the same, the surface of the orange; moreover, the way in which things look to you on both occasions is the same. So the objects absent in case 2 cannot be responsible for how things do look to you on that occasion. Whatever it is that is responsible in case 2 for determining the way things then look must be present on (p. 707) that occasion and be looking to one an appropriate way: we think things look as they do because we see the surface of the orange and the way it looks. But now, if that same thing is present in case 1 as well as in case 2, and there is no difference otherwise in how things look, then doesn't that show that it is the common object between case 2 and case 1, the surface of the orange, that is responsible for how things look in both situations? After all, ex hypothesi, the additional elements present in case 1 have made no difference to how things look to you given that the scene looks the same both in case 1 and in case 2. That suggests that it is how the surface of the orange is (and the background surfaces), and how it (and they) look to you, that constitute how everything in the scene looks to you. Although we say you see the orange in case 1, the orange itself, as opposed to its frontmost surface, does not play a constituting role in relation to your experience.

This seems to offer us the parallel for the case of vision to the role exclusion of bodies plays in relation to location. A primary role for objects of perception is to make up the array of elements that look some way to us. The immediate objects of sight will then be those that not only look some way, but which, in so looking, fix the way in which all objects we perceive look to us to be. Other objects will count as being seen (and hence as being seen mediately) through their relations to the objects immediately seen. It is through the immediate object of perception, and the way it looks, that these mediate objects come to look some way to us. The relation of part to whole seems to offer the relevant relation in the case of the orange and its frontmost surface, but there are other relations that seem capable of providing for the same. For example, the curtain moves and one sees the burglar enter the room. Perhaps in this case one does see the burglar. Yet matters could look just the same way if no burglar was there (perhaps there was just a sudden gust of wind). So even if you see the burglar in the actual case, the seeing of him does not determine the ways things now look to you. You see the burglar in virtue of seeing how the curtain looks, since across the two situations you see the curtain and in each it looks the same way. We can extend the line of thought further. There is a conceivable situation in which, through over-activity of moths, the curtain is no longer there except for the wisp of its foremost pile. Still, here matters can look the same to you as in the other two cases. So the curtain in turn is seen only in virtue of seeing its surface. This suggests that we might be able to provide an ordering of the objects of vision: immediately seen, seen only in virtue of a distinct, immediately seen object, seen only in virtue of a distinct object, seen only in virtue of another immediately seen object, and so on.

Our primitive idea of what it is for something to be seen, and for it to look a certain way, is for it to fix the way one then experiences, that is, the phenomenal nature of one's experience. Since we do say of other things that we see them too—we are prepared to say this of the whole orange in case 1, for example—we might surmise that, just as topological relations and political concerns can spread out the appropriate location of an object, so too some salient relations of belonging to, (p. 708) being a part of, or being a salient cause of, might play just such a role in the case of perception. There is, it must be said, a notable contrast here with the example of location inasmuch as the ways of specifying what belonging to should amount to as between immediate and mediate objects of perception is not at all clear, a fact that G. E. Moore was very sensitive to in his various discussions. That does not take away from the thought that the intuition driving one in the case of seeing rests with the condition for objects to be immediately visually perceived. And an object will count as an immediate object of perception where that object figures among those that determine the way one's experience currently is.

We started with the question whether one directly or immediately perceives any physical objects. Surfaces of objects are presumably themselves still physical. So the cases we have considered so far would not by themselves lead one to reject the claim that we immediately perceive physical things, even if we do not immediately see the ordinary objects around us that we have an interest in knowing about and interacting with. Still, it is easy to see how the terms of the distinction we have introduced could be extended to cover the proposal that one sees immediately only non-physical entities. Just as we considered in case 2 a situation in which the surface of the orange is present and seen as in case 1, and the manner in which things look is the same, so we might now consider a further situation—case 3. In case 3 not even the surface of the orange is seen. Case 3 is that of a hallucination indistinguishable from actually seeing the orange. Still, one might hypothesize, there is something seen in case 3 that is present in both cases 1 and 2 (just as the surface of the curtain is seen when the burglar is seen and only the curtain is seen): this would be some non-physical expanse that looks some way to the subject. Now, we might hypothesize as above that since the same thing is seen in case 3 as in cases 1 and 2, and how things look in all three is the same, then what constitutes the way in which things look to one in cases 1, 2, and 3 is the same: the presentation of the non-physical expanse and how it appears to be, which is the minimum that is present in case 3. So for this account of immediate seeing to provide the notion needed for the traditional debate, we must suppose that even when no physical objects are perceived, one can still count as seeing things in virtue of seeing some non-physical entities; and in addition we must hypothesize that these non-physical entities always play a role in visual perception, determining how things look to one even where one is also seeing a physical object.

Now this question, whether there are non-physical constituents of visual experience that determine how we visually experience things as being, is one that can be raised whether one appeals to the distinction between direct and indirect objects of perception or immediate and mediated seeing. Moreover, the rationale for claiming that we only immediately see non-physical entities can be understood perfectly well in these more general terms without drawing on the contrast between direct and indirect or immediate and mediate. The general form of those concerns will be (p. 709) addressed in the next section, but we should close this one by asking whether, even if we question the role of non-physical entities in visual perception, we must concede that we see objects only in virtue of seeing their surfaces.

3. ‘Now you see it; now you don't’: Seeing and Context-Sensitivity

Understanding the distinction between direct and indirect perception (or immediate and mediate perception) in this way does not yet give us the materials properly to interpret the traditional debate. Even if it were clear that one only saw the table before one through seeing a surface (or some surfaces) of the table, as we noted, this still does not show that one only sees the surface or the table because one sees some other non-physical thing. Yet before we even come to address the question of why one might think any non-physical object is also seen and plays a constitutive role in seeing physical things, one may already object to the contrast drawn, however simple and easy to apply it may appear.

The critic here may protest in terms introduced earlier, that it is through seeing things and thereby coming to be aware of them that we typically are placed to single those objects out in thought and come to know various facts about them. Perception plays a central role in our coming to refer to and know about things around us. But it plays this key role through making us aware of the objects and presenting them to us. The means of distinguishing between the immediate objects of perception and mediated ones requires us to recognize that only the immediate ones properly turn up in our experience of a scene and fix the way in which things as a whole appear to us. And this consequence, the critic complains, undermines our conception of how perception is useful to us. We take ourselves to be presented with such ordinary objects as tables and chairs, trees and dogs, cars and buses. But now, the critic continues, the reasoning rehearsed above seems to undermine the advantage we take such experience to give us. For while we start out with the idea that the orange itself is part of the experiential situation for us, when we reflect on case 2, we come to the conclusion that strictly speaking only the surface is an aspect of the experiential situation. We are therefore less privileged than we supposed ourselves to be, if the reasoning really is cogent. It hardly helps that in some other sense we still count as seeing the orange, albeit doing so indirectly, because that was not the sense in which we started out supposing that we stood in a relation to the orange.

(p. 710)

Suppose that one were moved by these concerns, and hence inclined to reject Jackson's suggestion that it is obvious that we see tables through just seeing their surfaces, or England through seeing the white cliffs of Dover. Where could one point to where the reasoning goes wrong? The most developed critique is offered by Thompson Clarke (1965).7 He argues that the discussion here ignores the context-sensitivity of talk of perceiving and appearing. The contrast between direct and indirect perception is a mistaken attempt to render into a context-insensitive vocabulary the irremediably context-sensitive issue of what one counts as perceiving on a given occasion. Given a suitable context and understanding of what is at stake, in staring at what is on the table, one does count as seeing the orange, even though given a different context and understanding the same situation is one in which one strictly only see the surface of the orange and not the orange itself. If one ignored the role that context plays here, we would simply have a clash of intuitions about what one sees: we both have the intuition that the orange is seen and the intuition that merely the surface is seen. Consistency could be restored by marking a context-independent distinction between immediate and mediate perception, but that would require that we have a context-independent way of fixing what one perceives or, as we saw in the last section, what is apparent to one and responsible for how the scene looks. The contextualist challenge then develops through insisting, first, that there is a prima facie clash in our intuitions and then that there is no context-insensitive way of resolving the clash.

To press this complaint, one needs to do two things. First, one needs to show that there is context-sensitivity that bears on the reasoning rehearsed in the last section. Secondly, one needs to give some reason to suppose that the context-sensitivity in question cannot be removed through suitable revision of the way we talk.

It is clear that what counts as seeing something can vary with context. Lost on an orienteering trip, you may spot a dot on the horizon and be correctly described as seeing the church on the hill—perhaps it is the only salient landmark in the area, and having noted it you are now able to use the map. By contrast, if you are traversing the moors inspired by reading a guide to English churches that announces that one's view of English Romanesque architecture will forever be changed on seeing the church, you will not yet count as having seen the church (i.e. your failure to be impressed as yet does not, as it stands, count against the guide's claim). So in the very same physical circumstances, you can both count and not count as seeing the church. But this example of context-sensitivity does not yet show that anything in the argument we rehearsed above trades on ignoring it.8

Better grounds for supposing that context-sensitivity in the original reasoning may be involved comes from reflecting on a tension among the claims in play in our original description of cases 1 and 2, from which we extracted the test for direct versus indirect perception. Consider again case 1: in this the orange is on the table (p. 711) before one, and we initially suppose that a proper description of this case is one of seeing the orange. Not only is the orange there on the table, but one can seemingly pay attention to it, pick it out, and wonder whether it would be good to eat, or a bit too dry or sour. We fix on case 2 as one in which the scene as a whole looks just the same to one (for it is by appeal to that thought that we reason that what is seen in both case 2 and case 1 is responsible for how things look, not only in 2 but also in 1). Yet, when we describe case 2, we are inclined to describe this case as one in which all that is apparent to one is just the surface of the orange. In this case the orange does not look some way to one. The surface may well look a certain way, the way such that it is as if there is an orange there, but there is no orange looking some way. Now one might suppose that still in case 2 we have a case of its looking as if there is an orange there when in fact none is present, much as in cases of hallucination. When caught by delirium it may look to a subject as if pink rats are running through the room, but we don't say that there are then any pink rats that look some way to that subject. Yet, we are also reluctant to assimilate case 2 to one of halluci-nation: in this case one does see all that there is to see; it is not as if some imaginary object occupies one's field of view.

So we now seem to have a tension among three claims concerning our descriptions of the two cases: (a) things look the same to one in both cases 1 and 2;(b) in case 2 solely the surface (and surrounding background) look some way to one; (c) in case 1 the orange and not (or not merely) the surface looks some way to one. No explicit contradiction has been demonstrated across (a) to (c), but still one may feel a tension between them. Isn't there a difference in the way things look in case 1 from case 2 if we say of case 1 that an orange is an element of how things look while finding no corresponding element to this in case 2?

I speak of tension here, rather than evident inconsistency, because the sense in which things look the same in cases 1 and 2 relates to the scene as a whole. But I have given no account of how the look of a scene should relate to how individual elements of a scene look to one. Inconsistency will follow immediately only if, from the fact that one scene looks exactly the same as another, it follows that there is nothing in one case that looks some way to the perceiver for which there is no corresponding element in the other situation. Once we admit that how a scene looks to one is a complex matter, involving different things looking some way or other to one, this principle no longer looks obvious. After all, it is conceivable that different elements in two scenes could combine so as to look the same overall. Even granting that, a tension remains: for there is an element in case 1, the orange looking some way, for which there is no element or combination of elements corresponding to it in case 2.

The tension would be removed if we could affirm simply that in both cases, strictly speaking, how things look to one is merely a matter of how the surface looks. And that, as we have seen above, is what the defender of the contrast between direct and indirect perception recommends to us. But we should be happy with this recommendation only to the extent that we are sure that it just isn't true of case 1 (p. 712) that the orange as such looks some way to one in this circumstance. This brings us back to the original complaint: that to accept this description of the experience is to construe it in a way involving less than initially we thought it did. It seemed to provide us with experiential access to the orange and not merely its surface.

The contextualist diagnoses the problem in a somewhat different way, and thereby rejects the use of the reasoning rehearsed in the last section. The contextualist will affirm the first claim (a) because there is a way of considering case 1 under which one counts as seeing the orange, and under which how things are presented to the subject includes that object. The contextualist will accept claim (b) because there is a context under which this is a correct description of how things are in case 2—obviously there is no context relative to which the agent perceives an orange in case 2 since there is no orange there to be seen, but whether the case counts as one in which what one seems to perceive matches with only what is there to be perceived is taken to be context-sensitive. We can construe case 2 as entirely veridical taking the subject to be liable to take in only what is there to be seen; but we can likewise construe it as misleading given its similarity to case 1.

For the contextualist, then, what the tension among the intuitions reveal is not that strictly speaking we don't have visual awareness of physical objects as opposed to their surfaces, but rather that we don't have a context-independent intuition about what one counts as either seeing or having a visual experience of. In the light of this we won't accept as compelling any form of the reasoning that keeps track of the contexts we are concerned with: in the context in which the orange counts as apparent to one in case 1, the appearance of things is not fixed in case 2 just by the way the objects present there look, for there is also the illusory aspect of the presence of the orange in addition. If we take instead the context in which in case 2 the way things look match what is there to be perceived, then there is an aspect of how things can look to be in case 1 that doesn't echo this, namely the perceived orange. This is not to say that it is wrong to suppose that there is a way in which there is nothing more to be seen in case 1 than what is there to be perceived in case 2;when one focuses on this way of thinking of case 2, one can construe what is apparent to one in case 1 also as merely the surface of the orange. What is mistaken, according to the contextualist, is to suppose from this that no truth can be expressed in saying that one sees the orange or that it looks some way to one.

This gives a sketch of how the context-sensitivity of ‘see’ or ‘appears to’ may bear on the cogency of the reasoning of the last section. But that is not yet to address the second, and equally important, concern. J. L. Austin complained that there are just different uses for the term ‘directly’ even in connection with verbs of perception, so there could be no one thing that philosophers could mean by indirect perception.9 In response, one may simply insist that one aims to introduce a specific and technical usage, such as that suggested by Jackson's definition for specific philosophical (p. 713) concerns. By analogy, one may respond to the contextualist's challenge that while it may be true that our ordinary usage of ‘see’ or ‘looks’ is context-sensitive in the ways sketched here, and no doubt in other ways too, still that does not show that we cannot introduce a special technical usage of ‘immediately see’ which, by stipulation, is intended to be context-independent. It is this that the theorist needs, and which the objector has not yet said anything against.

After all, it is familiar that when one person says ‘It is raining’ and another utters ‘It is not raining’, they need not be disagreeing. The first may be talking about the weather in Paris, the latter the state of play in London. Here, implicitly, there is a further parameter that affects the truth of the utterance. It is both a parameter that we can make explicit and which we can fix a specific value for, if, for example, our concern is solely with the weather in London. If the context-sensitivity that the objection appeals to above is anything like this, then we may suppose that there is some further parameter that we fix on in our ordinary use of ‘see’ or ‘look’. The theorist who wishes to introduce a constant and technical sense of these terms would then just need to fix on some relevant value and ensure that this value is held constant in all of our debates. The theorist need not deny that there may be an ordinary context of use under which it is true that we see oranges and that these are manifest to us in visual experience. What they should insist is that for the purposes of a philosophical account of seeing, we need to specify a particular value, and relative to that it is correct to say only that the surfaces of oranges, and not oranges themselves, are manifest to us in visual experience.

The objector may retort that this is not getting the point at all. For the response assumes that there simply is a determinate fact about how the subject is experiencing the situation, or how things look to him or her and what they can thereby take in about the situation. Although we may report on these experiential facts in different ways, depending on the relevant contextual understanding of ‘sees’, ‘looks’, or anyway the sentences containing these verbs, what is reported on is much the same fact in each case. But that was not the way that the original complaint developed; there was no assumption that there was a fixed or determinate way in which things must have been looking to the subject. When we describe the sense experience of subjects, we are describing what they take in about the world around them and thereby what is available for them to single out in thought and discover things about. Indeed, the complaint starts from the thought that if the orange is not part of how one experiences things to be and only the surface of the orange is, then one is not properly related to the orange through vision. Now, once the contextualist grants that there is a correct description of the experience under which the subject simply takes in the surface of the orange, then that suggests that when so considered the experiential facts are different: the subject takes in a different and more restricted range of the situation perceptually.10

(p. 714)

Since our conception of sense experience here, or so the contextualist may insist, is just a matter of what aspects of the environment the subject has awareness of, the correct moral to draw from this is that there are not context-independent determinate facts about what one sees or how one experiences things to be. So there would be nothing genuinely experiential for the technical usage to pick up on.

The idea that one's experience does not have a unique, determinate character sounds like a radical response to the initial and quite intuitive thought experiments. Whether or not one follows the contextualist down this route, the disagreement on this matter brings out one of the key points at issue, concerning the significance of sense perception or sense experience within one's broader cognitive economy. In part this takes us back to the traditional division between direct and indirect realism about perception. But it also takes us beyond that, since the underlying disagreement can be raised without having to worry about the contrast between immediate and mediate perception, and hence without having to settle the question whether or not our sense experience is determinate in character.

The contextualist is moved by the thought that our sense experience is the presentation to us of the ordinary objects around us in the world that we normally take ourselves to perceive, and not something less than that, be that the surfaces of these objects, or some non-physical entity.

Now the indirect realist can establish his or her case only if both the indirect–direct distinction can be applied to sense experience, and also the claim that we only perceive the ordinary objects of perception in virtue of being aware of non-physical entities. A common way of understanding this debate is to focus on the positive grounds for supposing that there are non-physical objects of awareness. But the contextualist rejection of the immediate–mediate distinction suggests a different way of understanding this debate: that what is at stake is whether we can make sense of how our sense experience in itself gives us access to the objects we ordinarily take ourselves to perceive and be in a position to think about through perceiving them. If our experience is merely of the surfaces of these objects or of other entities interestingly related to them, then, according to the contextualist conception, we are not related to these things as we took ourselves to be.

The more fundamental question is whether there is any reason to suppose that we just couldn't have experience in the way that the contextualist supposes that we do of the objects around us. The concerns so far raised about the intuitive distinction between immediate and mediate objects of perception cannot establish that result. For, as we have seen, there is as much plausibility to the idea that we do have visual experience of objects as to the idea that we merely experience their surfaces. But, as we shall see in the next section, the traditional concerns about illusion or hallucination can be put to the service of arguing that one couldn't be aware of mind-independent objects in the way the contextualist presupposes.

(p. 715) 4. The Argument from Hallucination

The commonest strategy for arguing that we never directly perceive physical objects is to appeal to some form of the argument from illusion or hallucination. That is, writers seek to show that in no case are we ever aware of physical entities as the perceived elements of our sense experience: first, by highlighting some salient cases of illusion, where one misperceives some entity, or by highlighting some cases of hallucination, where all agree one fails to perceive any physical object; then arguing of such cases that they do not constitute sense experience in which we directly perceive physical entities; and then to make a generalizing move and argue that if this holds of the illusions or hallucinations, it must hold of all of the sense experience we enjoy.

Of course, this is only to offer the schematic structure of the argument. And in fact there is no one argument that can be identified as the argument from illusion or hallucination. Rather, there is a whole menagerie of considerations that writers have appealed to; these vary in the examples of illusion or hallucination taken to be critical to the case and how the generalizing move is made from the test cases to all sense experience. In general, presentations of these arguments are enthymematic (at best), or plain invalid. Moreover, it is often difficult to discern what the needed extra premisses are that would render the argument valid, or what further support could be offered for the principles required once identified.

Here I will focus on just one variant of the argument from hallucination. First I will assume that the argument is intended in itself to have a purely negative purpose: to show that we do not directly perceive physical objects, or in fact, more narrowly, that what I shall call a naive realist conception of sense experience must be false. This negative conclusion is consistent with various different positive conceptions of sense experience. It is consistent, for example, with the view that our sense experience is the awareness of some non-physical entities, sense-data, whose manifest qualities determine what our experience is like. Equally it is consistent with the view that our sense experience is representational in character and does not involve any relation of awareness at all. The interest of the argument, though, lies in the significance of rejecting naive realism. For, if it is true that this is the account that best articulates our pre-theoretical conception of how our sense experience is, then we can construe the alternative theories as different attempts to account for this apparent character of sense experience, while avoiding the inconsistency that naive realism falls into.

What then is a naive realist conception of experience? In the last section I suggested that the contextualists like Clarke concerning the verb ‘see’ are moved by the thought that were the objects of perception not properly part of the perceptual situation from the subject's own point of view, one would be in a much worse position perceptually than we ordinarily take ourselves to be. For Clarke, it wouldn't be (p. 716) enough to respond that even if we don't immediately or directly perceive the orange, we can still see it indirectly or mediately. For Clarke's concern is that when initially we think of our perceptual situation it includes the orange, but when we consider the reasoning that leads to the contrast between immediate and mediate perception, we are left with the thought that strictly speaking we are only aware of the surface, and so at a disadvantage with respect to the position we initially took ourselves to be in.

Elaborating on this, we might suggest that we not only think of perceiving objects, seeing, feeling, hearing, or smelling, as relations to what is perceived, but also that the mental episode or experience involved in so perceiving is a relation. That is to say, first the naive realist supposes that what it is to have a sense experience is for one to be aware of some entity or entities: be that some object, event, or property instance, or other aspect of an object. So conceived, the occurrence of a sense experience is a relational occurrence or state involving both the subject and the entities of which the subject is thereby immediately or directly aware, and hence requiring the existence of both for the experience to occur. Secondly, the naive realist supposes that at least some of the entities to which we are so related in experience are the mind-independent entities around us that we take ourselves to perceive. At the moment I can look out of the window across a north London vista: for the naive realist, the rooftops and chimneys, and the bare trees and the clouds in view, may all be constituents of my current experience. These rooftops, chimneys, trees, and clouds do not depend for their very existence on my current awareness of them: they would remain were I simply to disappear, or less dramatically to close my eyes and no longer be aware of them. So, for the naive realist, my current visual perception of the scene before me is a relation: it is a relation of awareness, between me and the various physical and mind-independent entities in the world around me that I currently directly perceive.

We will return below to the question whether such a picture deserves the epithet ‘naive’: is this really the best articulation of how our sense experience seems to us to be before we engage in theoretical musing about it? What is of more concern for us now is why such an account could not be correct. One limited strategy for falsifying the account is to argue that various of the qualities that we are naively inclined to attribute to physical objects and which we take these objects to manifest in our experience of them are in fact not exemplified by any physical items. For example, many writers argue that colours as we experience them could not be instantiated by physical objects; or at least, even if instantiated by such objects, could not be perceived by us. If this thought is right, the red of the tiles opposite, which currently I take myself to be aware of, cannot be an example of some aspect of the tiles themselves featuring as part of my experience—because it is not them but something else that exemplifies the redness I am aware of. This strategy is limited in what it can show against the naive view, though. Even if the view of colour is correct and it is also true that no subject ever perceives a physical object as (p. 717) coloured, it still won't have been established that one is aware of no physical object in visual experience. To show that the subject is not directly aware of any physical object, it also needs to be established that the only entities one is aware of in visual experience actually do possess a colour. Doubting the physical reality of colour does not in itself settle this further point. In contrast, the inconsistency highlighted by the argument from hallucination would directly rule out the truth of naive realism with respect to any object of awareness, if the argument is sound.

The argument takes naive realism to be inconsistent with two further assumptions that in themselves seem plausible and are quite widely accepted: one that we might call experiential naturalism, that our sense experiences are themselves part of the natural causal order, subject to broadly physical and psychological causes; the other, the Common Kind Assumption, that whatever kind of mental, or more narrowly experiential, event occurs when one perceives, the very same kind of event could occur were one hallucinating. The first of these principles constrains what can be the case in regard to certain hallucinatory states, and the latter implies that this result must apply equally to the case of sense experience enjoyed when veridically perceiving.

Our starting point is with the idea of certain kinds of illusion or hallucination. Typically, when clinicians discuss hallucinations, they have in mind the kinds of delusion suffered by psychotics or extremely mentally disturbed individuals: someone might hear the voices of angels, see their loved ones as replaced by aliens, or suppose him- or herself covered by ants. There is no assumption in these discussions that the delusions or hallucinations such unfortunate people suffer are anything in character like what it would be really to perceive these circumstances. Typically delusion leads one to lose a grip on the nature of reality, and that may be a loss of understanding of what one's sense experience is like just as much as it is of what the world is like. When philosophers wish to stress the significance of hallucination or the argument from hallucination, what they typically have in mind, instead, is the possibility of the occurrence of mental episodes that lack the appropriately genealogy to be perceptions of objects in the world around us, but which from the subject's point of view need not necessarily be discernible from such perceptions. That is to say, for our current purposes we need to work solely with the idea of what one might call ‘perfect hallucinations’. Consider your current perception of the environment around you. Perhaps you are staring out at a late spring evening, or lying in summer grass, or sitting in a dusky office reading a philosophy paper. It is quite conceivable that there should be a situation in which you could not tell that things were not as they are now: so it might seem to you as if you were staring at an orange on the table, or taking in the smell of new-mown grass, even though unknown to you in that situation you were not doing so. Your perspective on the situation would not, in that situation, distinguish how things were from how they are now. Now we might say that how you are in that situation is a matter of having a sense experience that is not a case of perception.

(p. 718)

It seems both conceivable that such sense experience should occur—that is, that there is nothing in our way of conceiving phenomenal consciousness that should rule out the occurrence of such perfect hallucinations—and that they should be brought about through appropriate stimulation of a subject's brain. Experimenters have only been able to induce through direct stimulation of the cortex very simple visual phenomena—the flashes of light or phosphenes, for example, exploited in artificial visual systems for the blind.11 But this seems a merely practical or medical difficulty. So that we think, as embodied in the assumption of experiential naturalism, that our sense experiences, like other mental episodes, are subject to broadly physical causes.12 Once one has suitably manipulated the physical and psychological conditions, however, one has done all that can and need be done in order to induce in a subject a suitable sense experience.

But now suppose that we take hallucinatory experience to be relational in form, just as the naive realist takes veridical perception to be. There can be no instance of a relation without appropriate relata: so for the experience to occur, there needs not only to be a subject of the experience, but also the entities of which the subject is thereby aware. Since we are, ex hypothesi, dealing with a case of perfect hallucination, we know that there are no candidate physical objects of awareness. If you have the perfect hallucination of an orange, then the existence of this experience does not require that there be any real oranges, or other physical entities that look alike, to exist or to be appropriately situated for you to be aware of them. Nonetheless, if it is relational in nature, the experience can occur only if there is something or other for the subject to be aware of.

If we hypothesize (as the early sense-datum theorists did) that the relevant relata must be mind-independent, then the conditions for bringing about an experience must involve not only suitable manipulation of a subject's body and mind but also a guarantee of correlation between the perceiver and candidate objects of awareness. Given that such entities would be non-physical (since this is a hallucination) and non-mental (being awareness-independent), causing someone to have a hallucination would require causal influence over something neither physical nor psychological: the non-physical object of awareness. Therefore, if one holds to experiential naturalism and countenances only physical and psychological causation, one must deny that any such influence could take place. The only way, therefore, that the physical and psychological conditions could be sufficient for bringing about the experience is through them being sufficient for the existence of its object as well. And that can only be because the kind of occurrence one has brought about, the kind of state of awareness that occurs when one is hallucinating, (p. 719) is such that its occurrence constitutively guarantees the existence of an object of awareness.

That is to say: the reason that one might hypothesize that sense-data, or impressions, or percepts should be conceived as mind-dependent entities if they are to exist derives from the causal conditions for bringing about sense experience. It is only if the sense experience is itself constitutively sufficient for the existence of its object that one can assume that the physical and psychological causal conditions for bringing about sense experience are sufficient for bringing it about while also supposing such experience to be relational.

So far, this conclusion is not inconsistent with naive realism as I have characterized it. For that is a view about what we succeed in being aware of in cases of genuine perception. That theorist need have no pretensions about explaining how we can come to have sense experience when merely hallucinating, and so may well be agnostic about whether we are aware of anything at all when hallucinating, or aware instead of some mind-dependent entity. The naive realist need not take a view on whether merely mind-dependent entities could present the same appearance to a subject as a physical object. Ducks and decoy ducks can look the same while being very different in nature; why can't oranges and mental impressions have a common appearance too? Austin (1962) pressed just this question, and since he thought there was no good answer, he summarily dismissed the force of any argument in this area.

However, inconsistency does arise once we suppose that the very same kind of thing must be occurring when one has a perfect hallucination as when one perceives. Why? By the Common Kind Assumption, whatever kind of experience does occur when one perceives, the same kind of experience can be present when one is hallucinating. So if a hallucinatory experience must be of a kind that constitutes the existence of its objects, then since the very same kind of experience is also present when perceiving, that too will constitute the existence of its objects. That is, for any aspect of the perceptual experience the naive realist hypothesizes to be a relation to a mind-independent entity, consideration of the corresponding hallucination shows the entity in that case to be mind-dependent, and hence, any experience of that kind thereby to have a mind-dependent object rather than any mind-independent one.13 Mind-independent entities cannot, it follows, be constituents of the experience, contra the naive realist.

Suppose, then, that the naive realist seeks to avoid the conclusion by denying that any sense-data or other awareness-dependent entities are involved in hallucination. The alternative is to deny that the hallucination has any constituent elements. What account of hallucination is consistent with this denial? The commonest approach is (p. 720) to embrace a representationalist or intentionalist construal of experience. The denial that the experience has any constituent elements must be made consistent with the evident fact that, from the subject's perspective, it is as if there are various objects of awareness presented as being some way or other. That is to say, whenever one has a sense experience such as seemingly viewing an orange on the table, one's experience has a subject matter (as we might say)—there seemingly is a particular kind of scene presented to the subject in having the experience. And it looks as if the description of this subject matter carries with it a commitment to the existence of what the naive realist thinks of as the constituents of experience in the case of veridical perception. Since we deny that there are any such constituents of the experience in the hallucinatory case, our talk here must be lacking in ontological import. We are treating the hallucinatory experience as if it is the presentation of objects when in fact it is not. Intentional theories of experience take the description of the subject matter of an experience to express the representational or intentional content of the experiential state. The experience has its phenomenal character, according to this approach, in virtue of its possession of this content. In general we take ascriptions of representational content to psychological states to lack ontological commitment.14

Again, by the Common Kind Assumption, whatever kind of experience occurs when one perceives, that same kind of event will be present when one hallucinates. So if the hallucinatory experience lacks any constituents, then the perceptual experience, being of the same kind, does not have any constituents either. Although there may be objects that do act as appropriate values for our quantifiers, or referents for our terms, when we describe how things are presented as being to the subject of the perceptual state, none of these should be taken actually to be aspects of the experiential state itself, since such a kind of experience can occur when the subject is not perceiving. On this view, even in the case of veridical perception, when we make mention of the particular objects that the subject is perceiving, we do not describe them as parts of the experiential situation, but make mention of them to express the representational import of the experience. Given the naive realist's commitment to thinking of perceptual experience as genuinely relational between the subject and a mind-independent world, this representationalist construal of hallucination is no more amenable to naive realism than the sense-datum conception.

But what is the force of talking of a common kind here? There are ways of construing the Common Kind Assumption on which it is trivially false. If we relax our conception of a kind of event sufficiently, then any true description of an (p. 721) individual event introduces a kind of event. On such a conception, it is easy to find kinds that some individual events fall under and otherwise matching individuals fail to. You paint your picket fence white on Tuesday and I do so on Wednesday: mine is a Wednesday painting, yours a Tuesday one. Given the different descriptions, these seem to be different kinds of event. Since no party to the debate about perception denies that there are some descriptions true only of the perceptual scenario, namely that they are perceptions rather than hallucinations, someone who wants to take the Common Kind Assumption to be a significant addition to the debate cannot be using this broad conception of a kind of event.

For the Common Kind Assumption to be a genuine condition on the argument, and not simply a trivial falsehood, we need some conception of the privileged descriptions of experiences. For it to be a substantive matter that perceptions fail to be the same kind of mental episode as illusions or hallucinations, we need some characterizations of events that reflect their nature or what is most fundamentally true of them.15 I assume that we can make sense of the idea that there are some privileged classifications of individuals, both concrete objects and events, and that our talk of what is essential to a given individual tracks our understanding of the kinds of thing it is. That is, I assume the following: entities (both objects and events) can be classified by species and genus; for all such entities there is a most specific answer to the question ‘What is it?’16 In relation to the mental, and to perception in particular, I assume that for mental episodes or states there is a unique answer to this question that gives its most specific kind: it tells us what essentially the event or episode is. In being a member of this kind, it will thereby be a member of other, more generic kinds as well. It is not to be assumed that for any description true of a mental event, there is a corresponding kind under which the event falls. The Common Kind Assumption is then to be taken as making a claim about the most specific kind that a perceptual experience is, that events of that specific kind can also be hallucinations.17

(p. 722)

The background concern of the debate, therefore, is with the kind of occurrence that is involved in perceiving. When the naive realist insists that awareness is a form of relation and that we can bear it to mind-independent objects, this brings with it a commitment that relates to all of the kinds of cases in which such episodes can occur. Opponents appeal to the case of perfect hallucination not because they suppose that we can easily bring these states about, or because they think that in general illusions and hallucinations are just like perceptions, but because they want to test what the possible circumstances are for having this kind of awareness.

5. Indirect Realism Again

Experiential naturalism and the Common Kind Assumption taken together rule out naive realism about all aspects of sense experience. What bearing does that have on our original questions about direct perception and the role of sensing in our knowledge of the world? One way of responding to the argument is to hold on to the assumption that sense experience is a relation of awareness, but in the light of the constraints from experiential naturalism, suppose that the only objects of awareness are mind-dependent entities, sense-data, or impressions. If ordinary perception is constituted through one's sensory awareness, such a view will be close to, if not a version of, indirect realism about perception. Sense awareness will be constituted by awareness of mind-dependent entities, and through that one will count as perceiving (some of) the environmental causes of this state of awareness. The view will count strictly as a form of indirect realism, if one supposes that the sensuous awareness one has in having an experience is itself a mode of perceiving or sensing; for then one can apply the contrast between immediate and mediate perception introduced above.18 On the other hand, the theorist may as easily resist the claim that we perceive these objects of awareness—after all, our sense organs are not affected by them and we don't interact with them in the ways we do with the ordinary objects of perception. The theory will simply claim that what it takes to perceive objects at all is to have a sense experience and then go on to claim that such experience is the awareness of mind-dependent entities and qualities, even though not the perception of them.

On either construal, the original complaint about the position will have force. On initial reflection one's experience would seem to take in the ordinary objects in (p. 723) the environment around one. But, given reflection on the argument from hallucination above, one realizes that one's sense experience encompasses merely the mind-dependent entities constituted through one's awareness of them. Sense experience takes in less than it did, and hence seems to put one at a disadvantage relative to the position one initially took oneself to occupy.

Note that the worry here is not that one can only think about objects if one can genuinely perceive them, or that one can only know facts that one perceives; or that the objects of perception and the perceptually known facts must somehow be basic to all one's knowledge. Any such claims would be contentious and inconsistent with some familiar everyday examples: we know much about the world that we have learnt through testimony of books or other people; we know of and think about many things that we have never encountered. The worry here turns on a more modest point. When subjects first reflect on what they are in a position currently to find out about their situation and what they can think of, they take sense experience to involve the ordinary objects around them as available to attend to. As a consequence of reasoning through the argument from hallucination, they realize that that is not what sense experience provides them with. So, on one construal their thoughts are really about something different from what they supposed: one judges that that mind-dependent patch is an ink bottle, when supposing that the judgement was initially about the piece of glass five feet away. Alternatively, they do succeed in making the judgements that they supposed themselves to make, but they lack the understanding of how and why they made just that judgement that they originally started with. It seems obvious at first why it should have been that ink bottle that one picked out to think about: it was the one in view, salient within experience. In the light of the argument, however, we are to recognize that no such physical object is salient at all in experience.

6. Intentional Theories of Perception as Direct Realism

But this consequence follows not from the argument from hallucination alone, but from the further commitment that we must conceive of sense experience as relational in character: if not a relation to mind-independent objects like the tables, chairs, trees, or birds around us, then a relation to some entity, sense-datum, impression, or image that is conjured up in our awareness of it. The alternative response is to deny that sense experience has a relational character at all. For if there need be no object of awareness in the case of hallucinatory experience, there (p. 724) is no reason to suppose that there must be anything mind-dependent involved in sense perception at all.

Initially responses to sense-datum theories of sense experience sought to avoid this ontological commitment just by denying that sense experience is a relation to anything: so-called adverbialists about sense experience, for example, claim that to have a visual experience of a red patch is not thereby to stand in any relation to an object, some red thing, that one senses, but rather to have experience in a certain manner, to sense redly.19 But to put matters this way is to seek to avoid the onto-logical commitments of the sense-datum approach without necessarily addressing what leads the sense-datum theorist to posit the controversial entities in the first place. The sense-datum theorist posits the mind-dependent entities to play the role in sense experience that the naive realist supposes the ordinary objects of percep-tion to play: those things to which one has access in reflection on what it is like for one now to be perceiving or seeming to perceive. The adverbialist talk of sensing red merely being a manner of sensing does not obviously pick up on this element. Rather it would seem to deny that the sense-datum theorist is right to suppose that there is anything we are thereby aware of when we have sense experience, and hence to reject the naive realist's starting point as well.

A representational or intentional approach to perception, on the other hand, promises to offer an account that addresses these concerns.20 The minimum commitment of an intentional theory of perception is simply to attribute representational or intentional contents to sense experience: these states represent objects or one's environment as being a certain way, and the sense experience will count as veridical to the extent that how matters are represented coincides with how they are. Typically an intentionalist account of sense experience, however, seeks to account for the phenomenal character of such experience. Some suggest that this requires that the phenomenal character supervene on intentional content such that there can be no difference in character without a difference in intentional content.21 But in the current context a more substantive commitment is required: that that aspect of experience which the naive realist treats as a relation to the object of perception instead be treated as arising out of the representational or intentional properties of the experience. That is to say, where the naive realist supposes (p. 725) that to have the sense experience one stands in a relation to the object of perception, the intentionalist suggests instead that the experience represents the presence of the object and that this is not strictly a relation to the object in itself because the same kind of experience can occur with no object there. For the intentionalist, when one has a sense experience it is as if one is related to the objects of perception, as if the kind of awareness or attention to them requires the presence of those objects, but in fact experiences are such that this isn't so: for the same experience could occur and no objects be there.

The intentionalist does not deny the seeming role of the objects of perception in characterizing what the sense experience is like; he or she recognizes that the objects of perception will be part of the sense experience's subject matter. The only possible dispute with the naive realist is over what it takes for an object to be the subject matter of an experience. Therefore, one may conclude that this approach avoids the problems that the sense-datum theory raises. Intentionalism about perception, if it can consistently be fleshed out, seems to offer us a way of understanding how sense experience can be directed on mind-independent objects even given that experience can occur when we hallucinate. Intentionalism, after all, comes to be a way of defending a kind of direct realism about sense perception.

What, then, does the debate between the naive realist and such intentionalism amount to? What could show that sense experience of objects around us was or was not relational in form? One of the ways of thinking about this is to fix on the case of hallucination. In accepting the Common Kind Assumption, the intentionalist supposes that the same kind of thing can occur when one has a hallucination as a veridical perception. So there should be a correct description of your experience that can apply equally to the hallucinatory case and to the veridical perception.

If one is moved by the original thought, one might say that in the veridical case, the experiential situation should be describable in terms of the actual objects perceived: one is aware of the orange on the table and various of its features. If that is the correct description of the veridical case, then it is not directly available in the case of the hallucination given that no object is present there. So there needs to be a description of the hallucinatory case that avoids that commitment. One way, as Strawson (1979) stressed, is to describe it as a case in which it is as if one was presented with an orange. But what is the force of ‘as if’ here? If we take it as characterizing the hallucination by analogy with the case of veridical perception, then we will have to suppose that there is a distinct characterization to be given of the veridical case by which we can understand the comparison. But that cannot be the intention, since the presupposition is that whatever occurs in the veridical case is also present in the hallucinatory situation, so there should be some description in common that we can give directly of both.

To press this point further one needs to consider what the substance of the debate is between the naive realist insistence that sense experience must be thought of as a relation to the objects of perception, and the intentionalist's denial of this (p. 726) claim. At this stage, one might consider whether there is another way of responding to the argument from hallucination which preserves the naive realist view, namely a way of resisting the argument by rejecting one of its starting assumptions. For example, in different ways, transcendental idealists and early sense-datum theorists suppose that experiential naturalism is false. Merleau-Ponty (1942), for example, uses the causal argument to establish that our experience is not subject to the causal order and hence is outside the natural world, acting as limit on it. On the other hand, G. E. Moore (1957), C. D. Broad (1923, 1925, 1956), and H. H. Price (1932) all insisted that the immediate objects of awareness were non-physical but also not mind-dependent. At least for Moore, the initial impulse for this claim may have been that he wished to use sensing as an example of a mode of knowledge of a world independent of one's awareness of it. He takes it as definitional of knowledge that its objects are independent of the knowing of them, so if the argument from illusion establishes that the objects of sense are not the ordinary objects of perception, it cannot establish that they are mind-dependent. Apart from the oddities involved in supposing the world populated by non-physical entities with which we come into causal commerce only in perception, the resulting picture leaves the sense-datum theorist without a compelling way of showing why we must always be aware solely of such non-physical sense-data. So a naive realist might be untroubled by the hypothesis that hallucination involves the awareness of such strange non-physical entities.

If, though, the naive realist wishes to embrace a more conservative metaphysical picture and allow that we are just part of the normal natural, causal order and that we have no reason to suppose that the causal powers of this realm extend beyond the physical and psychological, then the only other option here would be to reject the Common Kind Assumption: that the very same kind of event that occurs when one perceives could have occurred were one hallucinating.

This approach to perception and appearance states is associated with so-called disjunctivism. The disjunctive theory of appearances (such labelling, I think, is due to Howard Robinson) was first propounded by Michael Hinton; the view was then defended further by Paul Snowdon and separately by John McDowell.22 In its simplest formulation, disjunctivism affirms that there is no highest common factor between perceiving and illusion or hallucination. Our description of our situation as one in which it seems to the subject as if there is an orange there covers diverse cases: the good case in which one sees the orange and is thereby aware of it as what it is; or a bad case in which this fails to be so, but in which it is as if the former obtains. What consequences would flow from denying the Common Kind Assumption? Does this give a coherent picture of sense experience?

(p. 727) 7. Disjunctivism about Appearances

In rejecting the Common Kind Assumption, the disjunctivist might be seeking to deny that there is anything really in common with respect to being an experience, or being a mental state, that perceptions, illusions, and hallucinations need have in common. This would be to deny even that the idea of a perceptual experience defines a proper mental kind, since all parties to the debate agree that this is a notion we can apply equally to veridical perceptions, illusions, and hallucinations. Yet given that disjunctivism seeks to defend naive realism, the rejection of the Common Kind Assumption only requires that one claim that the most specific kind of experience one enjoys when one perceives not occur when having an illusion or hallucination. This claim is the minimum needed to block the entailment from the claim that hallucinations cannot have mind-independent objects as constituents to the claim that the same is so of veridical perceptions. In this manner, the disjunctivist preserves naive realism through affirming

(I) No instance of the specific kind of experience I have now, when seeing the orange for what it is, could occur were I not to perceive such a mind-independent object as this.

and thereby denying the Common Kind Assumption.23

But what does this commit one to saying about the non-perceptual cases? At first sight, it may appear that all that the disjunctivist has to say is something entirely negative: that these are not cases of having the specific kind of experience one has when veridically perceiving. And hence one might think that disjunctivism avoids saying anything general about the nature of sense experience. In fact there is something more to say here that derives from what ought to be common ground to all parties to the debate.

Hinton began the debate about disjunctivism by focusing on a certain kind of locution, what he called ‘perception–illusion disjunctions’, for example, ‘Macbeth is seeing a dagger or under the illusion of so doing’.24 Hinton's strategy is to argue that there is no good reason to think that these disjunctive statements could not do all the work that our normal talk of appearances and experience do. That is, that there is no good reason from our ordinary ways of talking to suppose that we are committed to the existence of some special kind of experiential event that may be (p. 728) present equally in cases of perception and hallucination. Now this strategy prompts a question: Why pick on these disjunctions, then, rather than, say, ‘Either Macbeth is seeing a dagger, or he is under the illusion of seeing twenty-three pink elephants’? The answer, I take it, is that the disjunction Hinton highlights has the same evidential profile as self-ascriptions of perceptual experience. Someone in a position to make a warranted judgement about their experience can also put forward one of Hinton's perception–illusion disjunctions, but not so the alternative just suggested. One can gloss this, I suggest, by highlighting the connection between our talk of perceptual experience and the epistemic position a subject is in with respect to his or her perceptions and certain illusions or hallucinations, that they are indiscriminable from the perceptions through introspective reflection.

Above I suggested that philosophers’ use of the argument from hallucination focuses not on examples of actual hallucination as it occurs in dementia or delusions, but rather on the theoretical possibility of perfect hallucination. The disjunctivist can exploit just that idea to offer an alternative construal to the notion of sense experience implicit in the Common Kind Assumption. Suppose you start out only with the notion of veridical perception: what could introduce you to the idea of sensory experience more generally, to include illusion and hallucination? The idea of perfect hallucination sketched earlier was just that of a situation which from the subject's perspective is just like the veridical perception of a given scene, but which in fact is not a case of veridical perception at all. And surely it is at least cases like these that we have in mind when we think about examples of sensory experience that are not cases of veridical perception. We have a broader conception of sense experience than this, of course. For we allow that we can have illusions and hallucinations which are not veridical perceptions but that are not indiscriminable from perceptions: their character may vary wildly from what the corresponding perception would be like. But for the sake of this chapter, I want to work with the simplifying assumption that throughout we are to deal with what we might call perfect hallucinations. And for the case of perfect hallucinations, one could get someone to track the relevant cases in just the way suggested here.25

It is this idea, I suggest, that disjunctivists such as Hinton use in order to explicate their preferred notion of sense experience in general, i.e. that which generalizes across veridical perception, illusion, and hallucination. For in using this methodology, one can introduce, at least as a first approximation, the range of cases in dispute among the parties, without yet having to admit that there is something of the sort in common between perception, illusion, and hallucination of the sort (p. 729) that Hinton wishes to dispute. And hence this gives us the second commitment of disjunctivism:

(II) The notion of a visual experience of an orange on a table is that of a situation being indiscriminable through reflection from a veridical visual perception of an orange on a table as what it is.

We should immediately note three points about (II). First, the acceptability of (II) turns on how we are to understand the notion of indiscriminability. The relevant conception of what it is for one thing to be indiscriminable from another is that of not possibly knowing it to be distinct from the other.26 To be somewhat more pre-cise, since here we are concerned with knowing of individual experiences whether they are among the veridical perceptions or not, we can gloss it as:

¬◊ K[through reflection] ¬ x is one of the Vs.

(That is, x is such that it is not possible to know through reflection that it is not one of the veridical perceptions of an orange on the table as what it is.)27

This condition is met whenever x is one of the Vs, but if there are truths that are unknowable through reflection, then the condition can be met in other ways. It should be stressed that it is no part of this discussion that we can analyse or reduce the truths concerning indiscriminability, modal facts concerning the possibility or impossibility of certain knowledge, to claims about the sorting behaviour of individuals, or the functional organization that might underpin such behaviour. There are delicate questions for the disjunctivist concerning the link between a subject's failure to treat differently two situations and the claim that the two are indiscriminable for that subject.

(p. 730)

Secondly, the restriction ‘through reflection’ is an important and central addition here. When we describe the kind of case that fixes what we mean by perfect hallucination, we fix on one in which we unknowingly find ourselves in a situation that we can't know is not one of staring at an orange on the table. But we equally have a conception of sense experiences occurring where one has been tipped off about their non-perceptual status. If I take you into the bowels of William James Hall and subject you to an expensive visual-cortical stimulator so as to induce in you the hallucination of an orange, it seems quite conceivable that I should put you in a situation that in a certain respect is just like seeing an orange. In one important respect it is not: I have told you the experiment you will be subjected to. Since you have that information from my testimony, there is something you know that rules out your situation from being one in which you see the orange. Since we don't want to deny the possibility that this is a case of perfect hallucination, we need to bracket the relevance of the additional information you have acquired through testimony. This is what the appeal to ‘through reflection’ is intended to do. The situation in which you are knowingly having a hallucination of an orange is like a situation in which you don't know of the hallucination, because, if we bracket that additional information, then what is available to you otherwise, i.e. what is available to you in simply reflecting on your circumstances, does not discriminate between the two situations. The import of this restriction and the consequences that flow from it are central to understanding what disjunctivism is committed to, and how one should characterize one's objections to that picture of experience.

Thirdly, we should note that condition (II) just taken by itself ought to be interpretable as at least extensionally adequate on all theories of perceptual experience. Of course, the disjunctivist's opponent will not think that this properly gives an account of the nature of sense experience, and nor, for the matter, may it really articulate the concept or conception that we all have of what sense experience is. Nonetheless, the condition cannot fail to count as a sense experience anything that genuinely is one. For according to someone who accepts the Common Kind Assumption, the relevant condition for being an experience, being a P-event we might say,28 will be exemplified by both perceptions and perfect hallucinations. In both cases, then, the x in question will be one of the Vs, namely a P-event, and so it will not be possible for one to know that it is not one.29 The only way in which the extensions of our concept of sense experience and what is defined by (II) may fail to coincide is if (II) really is too liberal: that is, if it includes as instances of experience episodes that fail to be P-events. Now, as we shall see below, the full (p. 731) import of this possibility is a delicate matter. But at first sight, this is not a possibility that a theorist will wish to countenance. For after all, if in meeting (II) we describe a situation which from the subject's own perspective is just as if one is seeing an orange on the table, then how could it fail to count as a visual experience of an orange? For example, if the preferred account of experience is one in terms of sense-data, then this fact is not one entirely evident to us through initial reflection on our experience. As both intentional theorists of perception and naive realists insist, at least some objects of awareness are presented as the mind-independent objects of perception. There seems to be a privilege to the case of veridical perception in our description of any sense experience, whether it be a veridical perception, illusion, or hallucination (inasmuch as these cannot be told apart from veridical perception). Therefore, the theorist won't be able to isolate any feature that a P-event which isn't a veridical perception possesses and which the subject of the experience can tell through introspection that the event possesses but which a non-P-event which meets condition (II) can be known to lack. There would be nothing about the non-P-event that would rule it out from the subject's point of view as a candidate for being a sense experience in contrast to the P-event. Given this, someone who wishes to rule out such a case because it is not a P-event (whatever the particular account of experience is in question) seems to be offering us too restrictive an account of sense experience; for he or she seems to be interpreting what should at best be a sufficient condition for having a sense experience as a necessary condition. The catholicism of (II) in this case would suggest not that the account is too liberal in conditions on what is to count as experience, but rather that the theory in question (be it a sense-datum account, or some form of intentionalism) is just too restrictive in what it countenances as possible ways in which the kinds of sensory experience we have can be realized.

This suggests that the defender of the Common Kind Assumption should agree that there can be no case of one of us being in a situation indiscriminable through reflection from veridical perception, which is not a case of sense experience, whatever exactly the substantive account of sense experience the theorist thereby favours. The consequence of this is to accept certain constraints on the nature of sense experience and our knowledge of it. It is common for philosophers to suppose that conscious states must be (at least to self-conscious beings) self-intimating; such states will indicate their presence and some of their properties to the subject who is in them. What is required here is much more: that there should be no circumstance in which we are awake and there is no possibility for us to detect the absence of such states.

The disjunctivist's opponents need not reject (II) itself, or think of it as obviously implausible. They may even agree that our initial understanding of what sense experience is as (II) dictates, but then offer a more substantive account of what it takes for something to be an experience and so meet the condition in (II). On the other hand, they may think that the condition laid down in (II) itself is too thin, or (p. 732) modest, as an account of our understanding of sense experience. Still, for the reasons rehearsed above, they are unlikely to complain that (II) gets the extension of our concept of sense experience wrong. So (II) itself is unlikely to lead to any counter-intuitive consequences and on its own can hardly be considered a particularly controversial commitment of the disjunctivist. The same is not so, though, for the combination of (I) and (II). (I) commits us to thinking that there are some sense experiences that have a distinctive nature lacked by others, while (II) insists that all of these can nonetheless be indiscriminable from each other introspectively. Together this suggests that the phenomenal characters of two experiences can be different even while one of them is indiscriminable from the other. Many have supposed that what we mean by the phenomenal character of an experience is just that aspect of it that is introspectible, and hence that any two experiences that are introspectively indiscriminable must share their phenomenal characters, even if they differ in other ways.30

Now while some such complaint may have widespread support in discussions of phenomenal consciousness, it is not clear whether it should be taken as a primitive claim that is somehow obvious, and the rejection of which is incredible. After all, we can make at least some sense of the idea that distinct individuals, distinct events, and distinct scenes can all be perceptually presented to us and yet be perceptually indiscriminable from each other. That is, suppose that the individual experiences we have of the various individuals, events, and scenes we perceive thereby have as part of their phenomenal natures the presentation of those very objects; each of these individual experiences will be different from each other through featuring one object or event rather than another. Since distinct objects can be indiscriminable perceptually, it is plausible that these perceptions should be indiscriminable from each other introspectively. If so, distinct experiences will be different in ways that are not necessarily detectable through introspective reflection.31 It may be right in the end to dismiss such theories of perceptual experience as incorrect. But if there is an incoherence here, it is a subtle one, and not so glaringly obviously a contradiction. So this throws doubt on the idea that we should view the principle that sameness of phenomenal character is guaranteed by phenomenal indiscriminability as an evident truth. If we think the conjunction of (I) and (II) generates a counter-intuitive position, then there must be some further principle at work behind our thoughts that forces us to accept this strong condition.

Once one accepts that (I) and (II) are both true, then one must also deny that two experiences, one of which is indiscriminable from the other, must share (p. 733) phenomenal character (that is, one denies: any phenomenal character the one experience has, the other has too). It is consistent with accepting these two principles that one hold that such experiences would nonetheless share a phenomenal character. But a disjunctivist ought to reject even that claim, if the common phenomenal character is conceived of independently of (II), in terms of a positive characterization conceived independently of the unknowable difference across cases.

Why so? The problem that the argument from hallucination poses for naive realism relates to hallucinatory experiences that involve purely local causal conditions sufficient to bring about sense experience but which lack more distant conditions, such as the presence of a real orange in one's environment, which would be necessary to count the sense experience as a genuine perception of something in one's environment. The naive realist is motivated to reject the Common Kind Assumption because he or she insists that a necessary condition on having the experience he or she does when perceiving will be absent in such cases of halluci-nation. But in itself that doesn't determine what is necessary for one to be having a hallucinatory sense experience.

So now suppose that we manage to bring about a hallucination through repro-ducing the kinds of local causal conditions in someone's brain that otherwise result in the seeing of an orange. If no orange is in the environment, the subject is not seeing the orange, and according to the naive realist, is having an experience of a different kind from that had when perceiving. Still, we can ask, were the causes sufficient in this case to bring about a hallucination equally operative in the case of veridical perception? As we have set things up, this cannot be denied. So now it would appear that whatever occurs when one hallucinates (and the hallucination is brought about through the same local causal conditions as in veridical perception) also occurs when one veridically perceives.

This is strictly consistent with the Common Kind Assumption: one may both claim that the most specific kind of thing that occurs when one perceives does not also occur when one has a matching hallucination; and yet grant that whatever it is that occurs when one hallucinates, that same kind of thing does occur when one perceives. But still there seems to be a tension between the two commitments. Not least because one might otherwise suppose that what the disjunctivist accepts is the common element of perception and hallucination might by others be taken to be what is relevant to all of our explanations and accounts of sensory experience. Since everyone agrees that there is something about the perceptual situation not carried over to the hallucinatory one—it is, after all a case of perception of an orange on the table, while the hallucination is not—this threatens to reduce the disagreement to merely a verbal dispute. All sides agree that there is a common element present in both hallucination and perception, whatever is the kind of thing that can occur in both situations, and an element that is not carried over.

(p. 734)

The disjunctivist can block this manoeuvre by adding one further characterization of sense experience:

  • (III) For certain visual experiences as of an orange on the table, namely causally matching hallucinations, there is no more to the phenomenal character of such experiences than that of being indiscriminable from corresponding visual perceptions of an orange on the table for what it is.

Another way to put this point is to highlight that there are two sides to the disjunctivist's original conception of perception and sensory appearances. On the one hand is the thought that there is something special about the ‘good’ case, the presence of veridical perception and the apprehension of the mind-independent world. What holds essentially of the mental state or episode present in this case is not reduplicated across illusion and hallucination, so we can hold to the intuition that such states in themselves relate us to the mind-independent world. On the other hand, though, is the thought that in the ‘bad’ cases, the cases of illusion and hallucination, one is in a situation that fails to be the way that good cases are, but which purports to be the way that the good case is. Were a positive characterization always possible of the bad cases independent of their relation to veridical perception, were the notion of perceptual experience construable independent of this relation, then that these cases were bad would not be something intrinsic to them. This would not be a matter of us seemingly being related to the world but failing to be so, but rather being a certain way that we might also confuse with being perceptually related. So the disjunctivist thinks that there are cases of phenomenal consciousness that are essentially failures—they purport to relate us to the world while failing to do so. Commitment (III) makes this additional element clear in a way that (I) and (II) alone cannot.

But it does so through bringing out what many will take to be the high costs involved in endorsing a disjunctivist proposal. One can defend the naive realist conception of sense experience as applied to the case of veridical perception—that when one perceives, one is thereby in an experiential state that is a relation to the mind-independent objects of perception. The argument from hallucination can be blocked by denying that our conception of sense experience in general is that of a single kind of mental state that either is or is not a relation to something. Rather we conceive of sense experience in general as that which is indiscriminable through introspective reflection from perception. But in turn, when we reflect on certain kinds of case in which one has sense experience, cases of hallucination brought about through the same local causal circumstances as perception, we realize that the disjunctivist must go even further and claim that there is nothing more to having such an experience than being in a situation indiscriminable through reflection from sense experience: there is something necessarily delusive about the situation. Even if other accounts of sense experience really are unbelievable in relation to veridical perception, is the disjunctivist's conception of causally matching hallucination any more acceptable?

(p. 735) 8. Perceptual Contact

We started the discussion of this chapter with the idea that perception can relate us to the ordinary mind-independent objects we take ourselves to perceive, and that it is those objects that figure in our sense experience, where that is conceived as what is reflectively accessible to us when perceiving.

If this is construed in terms of naive realism, then the argument from hallucination sketched above, relying on experiential naturalism and the Common Kind Assumption, seems to reveal the falsity of this account of experience. And we can find one kind of motivation for traditional indirect realism in this: the sense-datum theorist embraces the idea that sense experience is relational in character, but in the light of the argument supposes that the objects of awareness must be mind-dependent and hence not the ordinary objects of perception.

An intentional or representational conception of sense experience would seem to promise a way of avoiding this result. For in denying that hallucinatory experience is a relation to any object, the approach can avoid making the claim that in having experience we must be aware of mind-dependent entities. So far, then, it seems as if an intentional or representational view could preserve the commonsense thought that sense perception makes available to us the objects of perception. But that requires that we can make sense of the idea that our perceptual experience, in being a representational state, may be as if a relation to something without really being relational in character.

In contrast to this, a disjunctive approach seeks to preserve the idea that sense experience really is a relation to the ordinary objects of perception in cases of veridical perception. To block the argument from hallucination it denies that we can be having the same kind of experience when hallucinating as when veridically perceiving. For reasons rehearsed above, this leads the theory to deny that there is any positive characterization to be given of certain hallucinatory states: nothing more can be said of them than that they are situations in which it is as if one is veridically perceiving.

This result is certainly one that many find counter-intuitive. But it should also remind us of the question raised earlier in relation to the intentional approach to perceptual consciousness. For that seeks to give a positive account of what hallucinatory experience is like in terms that are consistent with the thought that one is experiencing in just the same way when veridically perceiving and yet allow that veridical perception seemingly does relate us to the objects of perception present. However counter-intuitive the disjunctivist's position, a parallel problem seems to press on the intentionalist: how can we characterize the hallucinatory situation in the same way as that of veridical perception without undermining the thought that distinctively in the perceptual case alone one is aware of the objects of perception?

(p. 736)

If the traditional problem of perception is presented in terms of a contrast between direct and indirect perception, the persisting issue we face concerns our understanding of perceptual consciousness. The fundamental reason for resisting an indirect conception of perception is that the mind-independent objects of perception seem to us to feature in the subject matter of our sense experiences. Sense experience could only be a relation to objects in all cases of having experience if the objects in question are mind-dependent. The disjunctivist is moved by the thought that we do conceive of sense experience in the case of veridical perception as just such a relation to mind-independent objects. The cost of this view is to admit that there are other experiences for which a purely epistemological characterization must be given: there are cases simply of its being as if one is related to objects of perception, even when none are there. It might be thought that the intentional approach to perception avoids both this problem and the concerns with indirect realism. But it does so only if it can offer us a positive characterization of sense experience. And at that point the account seems to echo the disjunctivist's proposal that experience is as if one is perceiving.


Anscombe, G. E. M. (1965). ‘The Intentionality of Sensation: A Grammatical Feature’, in R. Butler (ed.), Analytic Philosophy, 2nd ser. Oxford: Blackwell.Find this resource:

    Armstrong, D. M. (1961). Perception and the Physical World. London: Routledge & Kegan Paul.Find this resource:

      ——(1968). A Materialist Theory of the Mind. London: Routledge.Find this resource:

        Austin, J. L. (1962). Sense and Sensibilia, ed. G. Warnock. Oxford: Clarendon Press.Find this resource:

          Baldwin, Tom (1990). G. E. Moore. London: Routledge.Find this resource:

            Broad, C. D. (1923). Scientific Thought. London: Routledge & Kegan Paul.Find this resource:

              ——(1925). The Mind and its Place in Nature. London: Kegan Paul.Find this resource:

                —— (1956). ‘Some Elementary Reflexions on Sense-Perception’, in R. Swartz (ed.), Perceiving, Sensing and Knowing. Berkeley and Los Angeles: University of California Press.Find this resource:

                  Byrne, Alex (2001). ‘Intentionalism Defended’. Philosophical Review, 110: 199–240.Find this resource:

                    Chisholm, R. (1959). Perception. Ithaca, NY: Cornell University Press.Find this resource:

                      Clarke, Thompson (1965). ‘Seeing Surfaces and Physical Objects’, in M. Black (ed.), Philosophy in America. London: George Allen.Find this resource:

                        Davidson, Donald (1969). ‘The Individuation of Events’, in Davidson, Essays on Actions and Events. Oxford: Clarendon Press.Find this resource:

                          Dobelle, W. H., Mladejovsky, M. G., and Girvin, J. P. (1974). ‘Artifical Vision for the Blind: Electrical Stimulation of Visual Cortex Offers Hope for a Functional Prosthesis’. Science, 183/123: 440–4.Find this resource:

                            Ducasse, C. J. (1942). ‘Moore's “Refutation of Idealism” ’, in P. A. Schilpp (ed.), The Philosophy of G. E. Moore. La Salle, Ill.: Open Court.Find this resource:

                              Fine, K. (1994). ‘Essence and Modality’, in J. Tomberlin (ed.), Philosophical Perspectives, 8. Atascadero, Calif.: Ridgeview.Find this resource:

                                (p. 737) Firth, R. (1965). ‘Sense-Data and the Percept Theory’, in R. Swartz (ed.), Perceiving, Sensing and Knowing. Berkeley and Los Angeles: University of California Press.Find this resource:

                                  Harman, G. (1990). ‘The Intrinsic Quality of Experience’, in J. Tomberlin (ed.), Philosophical Perspectives, 4. Atascadero, Calif.: Ridgeview.Find this resource:

                                    Hinton, J. M. (1967). ‘Visual Experiences’. Mind, 76: 217–27.Find this resource:

                                      ——(1973). Experiences: An Inquiry into Some Ambiguities. Oxford: Clarendon Press.Find this resource:

                                        Jackson, F. (1977). Perception: A Representative Theory. Cambridge: Cambridge University Press.Find this resource:

                                          Locke, Don (1967). Perception and our Knowledge of the External World. London: George Allen & Unwin.Find this resource:

                                            McDowell, J. (1982). ‘Criteria, Defeasibility and Knowledge’. Proceedings of the British Academy, 68: 455–79.Find this resource:

                                              ——(1984). De Re Senses. Philosophical Quarterly, 84; 34: 283–94.Find this resource:

                                                ——(1986). ‘Singular Thought and the Extent of Inner Space’, in P. Pettit and J. McDowell (eds.), Subject, Thought and Context. Oxford: Clarendon Press.Find this resource:

                                                  ——(1994). Mind and World. Cambridge, MA: Harvard University Press.Find this resource:

                                                    Martin, M. G. F. (1997). ‘The Reality of Appearances’, in M. Sainsbury (ed.), Thought and Ontology. Milan: FrancoAngeli.Find this resource:

                                                      ——(2002). ‘Particular Thoughts and Singular Thought’, in A. O'Hear (ed.), Logic, Thought, and Language. Cambridge: Cambridge University Press.Find this resource:

                                                        ——(2004). ‘The Limits of Self-Awareness’. Philosophical Studies, 120: 37–89.Find this resource:

                                                          Merleau-Ponty, M. (1942). La Structure de comportement. Paris: Presses Universitaires de France. Trans. A. Fisher as The Structure of Behavior (1963). Repr. Pittsburgh: Duquesne University Press, 1983.Find this resource:

                                                            Moore, G. E. (1922). ‘Some Judgments of Perception’, in Moore, Philosophical Studies. London: Routledge & Kegan Paul.Find this resource:

                                                              —— (1957). ‘Visual Sense-Data’, in C. Mace (ed.), British Philosophy in Mid-Century. London: George Allen & Unwin.Find this resource:

                                                                Neta, Ram (forthcoming). ‘Contextualism and a Puzzle about Seeing’. Philosophical Studies.Find this resource:

                                                                  Peacocke, C. A. B. (1983). Sense and Content. Oxford: Clarendon Press.Find this resource:

                                                                    ——(1990). ‘Perceptual Content’, in J. Almog, J. Perry, and H. Wettstein (eds.), Themes from Kaplan. New York: Oxford University Press.Find this resource:

                                                                      ——(1992). A Study of Concepts. Cambridge, Mass.: MIT Press.Find this resource:

                                                                        Pitcher, George (1977). ‘Berkeley’, in T. Honderich (ed.), Arguments of the Philosophers. London: Routledge & Kegan Paul.Find this resource:

                                                                          Price, H. H. (1932). Perception. London: Methuen.Find this resource:

                                                                            Robinson, H. (1985). ‘The General Form of the Argument for Berkeleian Idealism’, in J. Foster and H. Robinson (eds.), Essays on Berkeley: A Tercentennial Celebration. Oxford: Clarendon Press.Find this resource:

                                                                              ——(1994). Perception. London: Routledge.Find this resource:

                                                                                Searle, J. (1983). Intentionality. Cambridge: Cambridge University Press.Find this resource:

                                                                                  Sellars, Wilfrid (1968). Science and Metaphysics, ed. T. Honderich, International Library of Philosophy and Scientific Methodology. London: Routledge & Kegan Paul.Find this resource:

                                                                                    Snowdon, P. F. (1980–1). ‘Perception, Vision and Causation’. Proceedings of the Aristotelian Society, 81: 175–92.Find this resource:

                                                                                      ——(1990). ‘The Objects of Perceptual Experience’. Proceedings of the Aristotelian Society, suppl. vol., 64: 121–50.Find this resource:

                                                                                        (p. 738) Strawson, P. F. (1979). ‘Perception and its Objects’, in G. Macdonald (ed.), Perception and Identity. London: Macmillan.Find this resource:

                                                                                          Tehovnik et al. (2005). ‘Phosphene Induction and the Generation of Saccadic Eye Movements by Striate Cortex’. Journal of Neurophysiology, 93: 1–19.Find this resource:

                                                                                            Travis, Charles (2001). Unshadowed Thought. Cambridge, Mass.: Harvard University Press.Find this resource:

                                                                                              ——(2004). ‘Silence of the Senses’. Mind, 113/449: 57–94.Find this resource:

                                                                                                Tye, M. (1984). ‘The Adverbial Approach to Visual Experience’. Philosophical Review, 93: 195–225.Find this resource:

                                                                                                  —— (1992). ‘Visual Qualia and Visual Content’, in T. Crane (ed.), The Contents of Experience. Cambridge: Cambridge University Press.Find this resource:

                                                                                                    Valberg, J. J. (1992). The Puzzle of Experience. Oxford: Clarendon Press.Find this resource:

                                                                                                      Wiggins, David (1980). Sameness and Substance. Oxford: Blackwell.Find this resource:

                                                                                                        ——(1996). ‘Substance’, in A. C. Grayling (ed.), Philosophy: A Guide through the Subject. Oxford: Oxford University Press.Find this resource:

                                                                                                          Williams, Bernard (1978). Descartes: The Project of Pure Enquiry. Harmondsworth: Penguin.Find this resource:

                                                                                                            Williamson, Timothy (1990). Identity and Discrimination. Cambridge, Mass.: Blackwell.Find this resource:

                                                                                                              ——(2000). Knowledge and its Limits. New York: Oxford University Press.Find this resource:


                                                                                                                (1) (Jackson 1977, ch. 1). Although this work is a defence of a sense-datum theory of vision, Jackson more recently defends a form of intentionalism or representationalism about perception.

                                                                                                                (2) This he draws from Armstrong (1961), attempting to elucidate Berkeley. Cf. also Pitcher's interpretation of Berkeley on vision in Pitcher (1977: 9–13).

                                                                                                                (3) This he attaches to Price's discussion of the tomato. See the last two chapters for an alternative treatment of this passage. He also finds this in Don Locke's thought that immediate perception ‘does not go beyond what is perceived at the particular moment’ (Locke 1967: 171).

                                                                                                                (4) This he draws from Broad (1956) and Moore (1922).

                                                                                                                (5) So one might think that explanation is not really more basic here than our grip on ‘in virtue of ’. For one can as easily explain to someone that there are non-causal explanations by giving them cases in which they recognize that one thing holds in virtue of another but in which there can be no causal connection.

                                                                                                                (6) Whether one should suppose that this really is an aspect of our semantic competence concerning locative talk rather than part of our background understanding or theory of what it takes for medium-sized goods to be located is a more delicate question, however.

                                                                                                                (7) For discussions inspired by Clarke, see Neta (forthcoming) and Travis (2004).

                                                                                                                (8) Cf. Neta (forthcoming), who does not appear to note the gap.

                                                                                                                (9) See Austin (1962: 15).

                                                                                                                (10) The clearest development of this radical position is developed by Charles Travis in relation to thought content and representation in Travis (2001) and extended to the case of perception in Travis (2004).

                                                                                                                (11) Phosphenes, little flashes of light, were first investigated by Johannes Purkinje in the early nineteenth century. The effect can be produced just by physical pressure on the eye, but it has often been elicited through electrical stimulation. For its use in creating prosthetic vision systems, see Dobelle et al. (1974) and for a recent survey, Tehovnik et al. (2005).

                                                                                                                (12) And, remaining agnostic of the truth of dualism, psychological causes.

                                                                                                                (13) I assume here, in effect, that there cannot be constitutive over-determination of the veridical perceptual experience such that it is a relation to both the mind-dependent entity and the mind-independent one.

                                                                                                                (14) Or rather, more precisely, we may take the ascription to a psychological state of a given repres-entational content to lack the ontological commitment that assertion of that content (or of a propo-sition corresponding to that content if the content is non-conceptual or non-propositional in form) would involve. Some people, however, question whether one can avoid the ontological commitment inherent in the use of some referential terms in this way, cf. McDowell (1984). I assume that those drawn to intentional theories of perception will posit representational contents for perceptual states that avoid these difficulties. For more on this issue, see Martin (2002).

                                                                                                                (15) Note that this is not the same thing as to assume that the events we are interested in here are themselves part of the fundamental furniture of the universe. It is quite consistent with what is claimed here that there is a more fundamental level of reality out of which the mental is somehow constructed, or out of which it emerges. All that is rejected is that we explain the salience of this level of reality merely through appeal to an inclination on our part to describe some things as similar and others as different.

                                                                                                                (16) The most developed recent treatment of this kind of Aristotelianism about essence and nature is to be found in Wiggins (1980, 1996). For more on the question of essence, see Kit Fine's discussions of these matters (Fine 1994).

                                                                                                                (17) Can one formulate the argument, and the resistance to it, by avoiding mention of kinds? The argument from hallucination is often presented in terms of the causal conditions for bringing about a given instance of perceiving. That is, it is sometimes suggested that the issue turns on whether a given perception could have occurred without being a perception (cf. Valberg 1992). But there are many reasons for denying that the very same event could have occurred in a different causal context that have nothing to do with the debate about the nature of perception. (Consider Davidson's original criterion of identity for events: Davidson 1969.) If we do not assume that an individual event of hallucinating a picket fence is identical with a given perception, some additional principle must be appealed to in order to indicate that what is true of the one must be true of the other.

                                                                                                                (18) Moreover, contra the contextualist, the restriction of sense experience just to awareness of mind-dependent entities will not rest on ignoring the context-dependence of the words ‘look’ or ‘appear’.

                                                                                                                (19) Adverbialism is initially associated with Ducasse (1942), and later developed in very different ways by Chisholm (1959) and Sellars (1968) (it is doubtful whether Sellars's account does raise the question pressed in the text); and more recently defended by Tye (1984) against Jackson's (1977) critique. Tye has subsequently endorsed a form of intentionalism about perception.

                                                                                                                (20) Recent defenders of intentionalism include Harman (1990), Peacocke (1983, 1990, 1992), Searle (1983), Tye (1992). In the analytic tradition its popularity can be traced back to Firth's (1965) discussion of the percept theory, on the one hand, and Anscombe's (1965) critique of both sense-datum theorists and their ordinary language opponents, on the other. With some caveats, one can also see it as dominant within the phenomenological tradition. Armstrong's inclination to believe the theory may also be thought of as part of the tradition, although he hesitated to treat it as a form of direct realism about perception; see e.g. Armstrong (1968, ch. 10).

                                                                                                                (21) Cf. Byrne (2001).

                                                                                                                (22) See Hinton (1967, 1973); Snowdon (1980–1, 1990); McDowell (1982, 1986, 1994). There are significant differences in the formulation and motivation for each of these approaches. I discuss a little of this in Martin (2004). Robinson's critical discussion can be found initially in Robinson (1985) and is further elaborated in Robinson (1994).

                                                                                                                (23) As should already be clear from the naive realist commitment to having entities as constituents of perceptual episodes, the disjunctivist must reject any kind of physicalism that identifies kinds of mental episode with kinds of physical events in the subject's brain. In rejecting the Common Kind Assumption, the disjunctivist does not take a stance on whether the very same kind of local physical conditions can accompany veridical perception and hallucination.

                                                                                                                (24) See the works cited above in n. 20.

                                                                                                                (25) For a (too brief) discussion of how we can generalize away from the case of perfect hallucination to cover illusions and hallucinations more generally, see Martin (2004: 80–1).

                                                                                                                (26) The most extensive discussion of indiscriminability is to be found in Williamson (1990). Williamson principally focuses on the case of knowledge or lack of knowledge of identities and distinctness, that x = y or xy. As I note in the text, we are concerned with the plural form of whether x is one of the Vs. This form, even more obviously than the case of individual identities and distinctness, raises questions about intensional versus extensional formulations.

                                                                                                                (27) Jim Pryor and others have suggested to me that in our normal usage of ‘phenomenally indiscriminable’ this phrase should not be interpreted according to the above schema. The schema is not symmetrical: that hallucinating is not discriminable through reflection from perceiving does not entail that perceiving is indiscriminable from hallucinating (cf. Williams 1978, app.; Williamson 2000, ch. 6). But, the complaint goes, it is just obvious that as we use talk of ‘phenomenally indiscriminable’, this relation is symmetrical. In response, I would suggest that we should be more respectful of the etymology of the term, which would support the more complex form suggested in the text. That this should lead to a symmetrical relation in the case of phenomenal states is readily explicable without supposing it analytic of the notion. For the vast majority of philosophers in this debate do make further substantive assumptions about the nature of psychological states that would allow experiential states to be indiscriminable in our sense only if they are identical in phenomenal character. And it is just these substantive assumptions that the disjunctivist challenges.

                                                                                                                (28) That is, for example, an event of being aware of an array of sense-data with such-and-such characteristics, or an event of being in a state of mind with such-and-such representational properties or content.

                                                                                                                (29) Note also that, as formulated, (II) takes no stance on whether perceptions ever occur, or whether a subject need believe themselves ever to have perceived anything. All that it requires is that we accept that sense experiences have the character at least of seeming to be perceptions.

                                                                                                                (30) In effect, this is to press what I called principle (IND) in Martin (1997): ‘If two experiences are indistinguishable for the subject of them then the two experiences are of the same conscious character’ (p. 81).

                                                                                                                (31) I discuss this option for an intentional theory of perception in Martin (2002). There are delicate questions to be raised here about the interrelation between the phenomenology of individual experiences and the ways in which experiences are similar or different from each other.