chapter 4, 'Availableness and Occurrentness', pp. 60-87.

196

Perceptual Intentionality, Attention and Consciousness thought; they are articulated for me, for the purposes of action (it is for this reason that Heidegger is right to speak of a kind of intentionality here). Thirdly, when I direct my attention to the hammer or various objects in my surroundings it is, in the first instance, to the hammer and objects thus articulated and present. I suggest that much of what we mean by attention-guiding presence in everyday perception is on the level of Heidegger's notion of absorbed intentionality. A lot of what he says about it and theoretical or thematic intentionality is captured in John Campbell's distinction between causally indexical and non-indexical content. The causally indexical contents of perceptions are contents the causal and spatial significance of which is exhausted in terms of their implications for the subject's actions. The non-indexical contents are contents the grasp of which requires grasp of a primitive theory of perception, and the attendant conception of oneself as one object among others. Such a conception is, in turn, described by Campbell as being essential for theoretical as opposed to purely practical knowledge.18 Getting right the account of the relation between phenomenal and conceptual ingredients in perceptual intentionality is, then, partly, a matter of getting right the way in which such causally indexical immersed articulations of the environment inform and guide attention-mediated conceptual thought about it. 2. In the paper quoted earlier by Kahneman and Henik, Gestalt groupings and figure ground segregations are said to 'define the frustrating boundaries of what we can will ourselves to do' (in selectively attending to the environment).19 And, although not always put in exactly these terms, many of the extensive studies on the workings of selective and divided attention are in fact just studies of the constraints imposed by various organizational phenomena on the possibilities of divided and selective attention. Some quick examples will give a flavour of the kind of connection being talked about. The term 'selective attention' refers to subjects' capacity to isolate an object or property from others presented to them. An example of such failure is this. Suppose you are looking at a picture of a Rubin vase, which can appear either as a vase or as two profiles facing each other, depending on what is taken as figure and what is taken as ground. Suppose you see it as two profiles. Your task is to attend selectively to the contours of one of the profiles. You will not succeed. You will not recognize it if presented on its own, and certainly not if it represented as enclosing some other figure, say a vase. This is seen as showing that Gestaltists' claims about contours being treated as essentially belonging to the objects they enclose 18 Campbell, 'The Role of Physical Objects', pp. 82-8. 19 Kahneman and Henik, 'Perceptual Organization', p. 201. 197

Naomi Eilan reveal constraints on what we can selectively attend to in percep- i tion.20 (There is no comparable difficulty in thought of the contour on its own.) A famous example of the effects of Gestalt organizational phenomena is the Stroop test. If your task is to name the colour in which a word is printed, say red, but the word printed is, say, 'green', there will be interference: you will say 'green' instead of 'red'. But if the word 'green' appears very near the target word, there is no interference. This is taken to show the effects of Gestalt organization into objects on the working of selective attention. If properties are perceived as belonging to a single object they are hard to attend to selectively.21 A simplified example of Gestalt-imposed constraints on divided attention is this. Suppose your task is to report the colour of two dots. If they are seen as contained within a single object, as defined by Gestalt laws, you will be as good at that as in reporting the colour of one. But you will be significantly slowed down if the two dots are presented as being in two separate objects, as defined by Gestalt laws, even though the spatial separation is exactly the same as when they are presented as being within one object.22 The examples of immersed intentionality and the effects of Gestalt organization give us two distinct ways in which there is in perceptual experience a presence of the world to the subject, in contrast to thought and propositional memory. In the immersed case, what we have is an articulation of a world for the purposes of action which yields a presence of the world which is prior to attention and to which we may attend when we attend. But as we attend to it, the way it was present to us prior to attention disappears from view: we switch to a theoretical conception of the object. The kind of presence brought to light by examining the effects of Gestalt phenom20 E. Rubin, 'Figure and Ground', in Readings in Perception, ed. and trans: D. C. Beardslee and M. Wertheimer (New York: Van Nostrand, 1958), pp. 194—203. For an excellent account of Rubin's own work, and reports on experiments designed to update his findings and to give them computational underpinnings, see Jon Driver and Gordon Baylis, 'EdgeAssignment and Figure-Ground Segmentation in Short Term Visual Matching', Cognitive Psychology 31 (1996), 248-306. 21 See Kahneman and Henik, 'Perceptual Organization', and also for a wide range of other examples, including auditory ones. Also A. W. Garner, 'The Analysis of Unanalysed Perceptions', in Perceptual Organization, ed. Kubovy and Pomerantz, pp. 141—80. 22 See e.g. John Duncan, 'Selective Attention and the Organisation of Visual Information', Journal of Experimental Psychology: General (1984), 113, 501-17; G. Baylis and J. Driver, 'Visual Attention and Objects: Evidence for Hierarchical Coding of Location', Journal of Experimental Psychology: Human Perception and Performance 21 (1995), 1323-42. 198

Perceptual Intentionality, Attention and Consciousness ena is quite different. Here what we are shown are non-conceptual aspects of perceptual experience that are present as we attend to objects and which are partly determinant of the distinctive phenomenology of attention-mediated, perception-based thought in contrast to that of pure thought. There are many other ways in which extra-conceptual ingredients come into an account of perceptual presence, but these two will suffice for raising the final set of problems I will be discussing. The claim has been that the problem of getting perceptual intentionality right is the problem of showing how conceptual and phenomenal aspects are interwoven in perceptual intentionality. This in turn was said to lie in getting right the mixture of activity and passivity in perceptual experience. The mixture of activity and passivity we have now been discussing can be put as follows. Both immersed articulation of the environment and the Gestalt organizations in some way constrain the operation of attention. This constraint introduces a kind of passivity, lacking in pure thought. The problem of getting perceptual intentionality right is, in part, the problem of how we should explain and fill out this constraining relation between such non-conceptual, presence-yielding aspects of experience, on the one hand, and attention, conceived of as the vehicle of reason and concepts, on the other. Neither the immersed articulation of the environment for the purposes of action nor Gestalt organization of the environment into groups and objects provides reasons for attention-mediated thought about it. The relation of constraint is, then, not rational. Nor is it correct to speak of them as merely causing the attention-mediated thought. Constraint is not a purely casual relation. At this point the right thing to do would seem to be to reach for what was earlier said, in describing the first strand of mixture of activity and passivity in perception, namely that we should conceive of attention as active selection of information for further processing. What we might say here is that this account will yield, at the same time, the account we need of the relation of constraint. More specifically: both Gestalt groupings and object segregations, and the articulation of the environment for the purposes of action involve some way of representing the environment and should therefore have some kind of computational underpinning. What is needed for explaining the constraint imposed by them on attention is that we treat them as the information selected for further processing. We should then go on to say that they just are the inputs to the information processing involved once active selection has occurred. The relation of constraint is a computational one. This seems the right way to go; but I now come to what seems to 199

Naomi Eilan me to be a difficulty with this picture, however attractive, which is, in turn, a difficulty that I think any account of perceptual intentionally must face, in some form or other. The notion of a 'preattentive' stage in perception (the phrase was coined by Neisser23), is used by psychologists to describe information available in perception prior to, and independently of, attentional processing. If we are guided by phenomenological constraints only we will apply this notion both to the articulation of the environment for the purposes of action (in a way that accords with the spirit of how Neisser himself originally used it) and to the Gestalt groupings and object segregations (as Kahneman and many others use it.) Used in this way, the claim would be that an account of the mechanisms of perceptual intentionality just is an account of the way in which such presence-yielding preattentive information serves as input to the rational activity of selecting preattentive information for further processing. There is, however, a different use of the notion of a preattentive stage in perception which is guided not by phenomenological constraints but, rather, by computational constraints only. Here, preattentive information, for example in vision, is identified with whatever early vision delivers, in parallel, bottom-up. This is then conceived of as input to top-down operations, for example the kind of visual routines proposed by Shimon Ullman for the representation of spatial relations and shape properties.24 A problem in the way I have set out the relation between phenomenally determined preattentive information and attention is that there is no guarantee that the two ways of describing the preattentive level will coincide. In fact, there are numerous examples of them not coinciding as things currently stand. One example is this: if we use 'preattentive' to describe the articulation of the environment for the purposes of action then, according to Ullman's account of visual routines, we will have to say that such visual routines go into this kind of articulation.25 So what is preattentive phenomenologically speaking is not what is preattentive computationally speaking. More generally: on the going account, preattentive vision delivers features (rather than object representations) which are then combined in top-down processes to yield representations of objects. But as we have appealed both to Gestalt organizations and 23 Ulric Neisser, Cognitive Psychology (New York: Appleton-Century Crofts, 1967). 24 Shimon Ullman, 'Visual Routines', in Visual Cognition, ed. S. Pinker (Cambridge, MA: MIT Press, 1985), pp. 97-160. 25 The need for the use of such routines for the purposes of action is made explicit by Ullman, in ibid., p. 98. 200

Perceptual Intentionality, Attention and Consciousness to the articulation of the environment for the purposes of action, this was in order to describe a kind of preattentive presence of whole objects to the subject, rather than disconnected features. This (more than merely) possible divergence might appear to present us with a dilemma, neither horn of which is obviously attractive. One option is simply to abandon the link with the phenomenally motivated account of the preattentive level. But if the arguments presented in the first section of this paper are right, we cannot do that while simultaneously holding onto the idea that attention is the means by which rational activity and thought get into perceptual experience. Recall, the argument was: neither the blindseers' nor the superblindseers' perceptions provide them with reasons for judgement because of the absence of whatever is needed to explain phenomenal presence. Rationality here is a function of the link with such phenomenal presence. Drop the commitment to getting presence right and you forfeit your right to be treating attention as the vehicle by means of which rationality gets into experience in such a way as to provide experience with its epistemic role. The other option is to hold onto the link with phenomenology and to drop the attempt to link up with information processing stories. There are two reasons why this is not attractive, both connected with getting right the mixture of activity and passivity distinctive of perceptual experience. First, it leaves us with no way of giving a plausible, non-metaphorical articulation of what we mean by preattentive presence constraining the operation of attention. Secondly, the appeal to information-processing was, I argued, the only way to do non-metaphorical justice to the kind of mixture of activity and passivity common to both remembering and perceiving. The hardness of the problem of consciousness is usually said to stem from a clash between a general thesis to the effect that everything should, ultimately, have a scientific, objective explanation, and arguments to the effect that phenomenal aspects of experience necessarily resist capture in the objective net. It might be tempting to view the above dilemma as presenting us with a new version of this problem, incorporated here into accounts of 'consciousness of and access-consciousness in the case of perception. But this would be a mistake, for two reasons. First, to say there is no guarantee that phenomenally driven and computationally driven descriptions of the preattentive level will coincide, is not to say anything about whether we can give an account of what phenomenal presence is in purely information-processing terms. That latter is what concerns those preoccupied with the 'hard' problem of phenomenal consciousness, an issue which I noted at the outset I would not be addressing in this paper. The claim has been only that if the need to 201

Naomi Eilan account for phenomenal aspects of consciousness is what makes a concept hard then 'consciousness of and access-consciousness are hard concepts, at least in the case of perception. Secondly, and more importantly, to say there is no guarantee of coincidence is no resting place; and it is the fact that it is no resting place which introduces a different kind of hardness here. There are various moves yet to be explored and it is, to my mind, an open question whether and how they might work. For example: it might be possible to motivate a distinction between attention as it is used in the articulation of the environment for the purposes of action and attention as it is used in the articulation of the environment for the purposes of theoretical thought about it; and, relative to that, to motivate a distinction between two kinds of preattentive level such that what is preattentive for thought is not so for action. More generally, it may be possible to motivate a distinction between the use of attention pre and post the level of object representations, and, again, relative to that to introduce a distinction between two kinds of preattentive level. Moreover our phenomenological intuitions are hardly sacrosanct here: it is not as if we know in advance what should go into an account of presence. What is needed here is detailed open-ended investigation of the relation between psychologically and philosophically guided formulations of the nature of attention and its relation to consciousness and thought. The hardness here consists precisely in this open-endedness, in the fact that a priori arguments about the relation between science and phenomenology offer us no guidance as to how to proceed. 202

Experience and Reason in Perception BILL BREWER The question I am interested in is this. What exactly is the role of conscious experience in the acquisition of knowledge on the basis of perception? The problem here, as I see it, is to solve simultaneously for the nature of this experience, and its role in acquiring and sustaining the relevant beliefs, in such a way as to vindicate what I regard as an undeniable datum, that perception is a basic source of knowledge about the mind-independent world, in a sense of 'basic' which is also to be elucidated. I shall sketch the way in which I think that this should be done. In section I, I argue that perceptual experiences must provide reasons for empirical beliefs. In section II, I explain how they do so. My thesis is that a correct account of the sense in which perceptual experiences are experiences of mind-independent things is itself an account of the way in which they provide peculiarly basic reasons for beliefs about the world around the perceiver. Why must perceptual experiences provide reasons for empirical beliefs? I shall argue that their doing so is crucial to the determination of specific contents for such beliefs. In other words, I shall argue that unless perceptual experiences provide reasons for empirical beliefs, there can be no genuine beliefs about the mind-independent spatial world.1 This obviously needs clarification. Which experiences are Many thanks to John Campbell, David Charles, Bill Child, Naomi Eilan, Marcus Giaquinto, Peter Milne, Adrian Moore, Christopher Peacocke, Carolyn Price, Johannes Roessler, Tom Stoneham, Peter Sullivan, Rowland Stout, Charles Travis, Ralph Walker and Tim Williamson, for their helpful comments on previous versions of this material. My views on these issues have also been greatly influenced by as yet unpublished written work by John Campbell and Naomi Eilan. Work for this paper was supported by Research Leave funded by the British Academy. 1 This claim is a crucial component of J. McDowell's position in his Mind and World (Cambridge, MA: Harvard University Press, 1994). Indeed, the argument I offer in support of it is my own extended development of his very suggestive comments on the matter (see especially Lecture I). Both mind-independence and spatiality are crucial to the claim. Arithmetical 203

Bill Brewer claimed to provide reasons for which empirical beliefs; what exactly are reasons in this sense; and in what way do perceptual experiences provide them? Roughly, the relevant empirical beliefs are those with contents which it is possible to come to know directly on the basis of perception; and the corresponding perceptual experiences are those which would be involved in the acquisition of such knowledge. Equally roughly, reasons for beliefs are features of the overall set of circumstances a person finds himself in - by which I mean to include both his mental condition and its wider context, perhaps over a significant amount of time - which make it reasonable from his point of view to come to have or retain the beliefs in question in those circumstances. Furthermore, as reasons from his point of view, they figure as such in the subject's reflective thinking about his situation. In some way or another, he recognizes these features of his circumstances as the reasons which they are. Only so can they actually succeed in moving him in such a way that it is appropriate to cite their status as reasons in an explanation of his having the beliefs. Finally, perceptual experiences provide such reasons in the sense that they are essential to his entertaining certain demonstrative contents, simply grasping which constitutes a prima facie reason for endorsing them in belief. In outline, my argument is this. It is only in virtue of their relations with his perceptual experiences - that is, the immediate impact of mind-independent reality upon his conscious mental life - that a person's beliefs acquire genuine empirical content. These relations contribute essentially to fixing a given belief as a belief about a particular mind-independent thing to the effect that it is determinately thus and so. Yet if his experiences give him no more reason to believe that things out there are one specific way rather than any other, then they cannot possibly fulfil this role. His beliefs therefore fail to be beliefs about the mind-independent spatial world at all. So, unless perceptual experiences provide reasons for empirical beliefs, there can be no such beliefs. Two premises can be distinguished here. The first insists that beliefs concern mind-independent spatial reality only in virtue of standing in certain relations with perceptual experiences; the second thought, for example, even though its subject matter may correctly be conceived as mind-independent, is clearly a different issue. Conversely, I am concerned with our knowledge of the empirical world in a sense in which this is incompatible with any attempted idealist understanding of spatial particulars as, ultimately, mind-dependent. See J. Foster, The Case for Idealism (London: Routledge, 1982), chapter 5, for a development of this idealist alternative. To avoid undue repetition in what follows, though, I shall often use just one of the two adjectives as shorthand for the pair. 204

Experience and Reason in Perception claims that only reason-giving relations between experiences and beliefs will do here. I shall consider each of these in turn. Belief and Experience Perceptual experience is the world's direct impact upon a person's mind. It is the only immediate difference which is made to his mental life by his being the particular person which he is, tracing the particular spatio-temporal route which he does through the world. So, unless his beliefs about the world are systematically related in some way to this experience, they are utterly insensitive to his actual environment. Even if his world-view somehow survives this 'confinement' as a series of quasi-rational manipulations - the most abstract imaginable algebra perhaps - mind-independent reality drops out as quite irrelevant to whatever residual norms may govern it. It therefore fails to be a world-view, or set of beliefs about that reality, at all.2 Thus, beliefs concern mind-independent reality only in virtue of standing in certain relations with perceptual experiences. 1 should stress that this requirement is weaker than the quite general empiricist claim that every concept either has its source in experience, or is composed exclusively from simple concepts which do. It is certainly weaker than the extreme verificationist idea that the significance of any empirical belief whatsoever is exhausted by its association with a set of experiences which conclusively verify it. Rather, it simply insists that without some anchoring to his particular worldly environment through some relations or other which they bear to his actual or possible perceptual experiences, certain of a person's purported beliefs about the way things are in the world around him collapse into an empty game. All that this premise really amounts to is a denial of the extreme rationalist suggestion that such beliefs simply sit there in a person's mind, with their determinate, empirical, contents engraved upon them, quite independently of any relations which they may or may not have with the actual things around him through the impact of worldly affairs upon his conscious experiential life. This first premise appears to face a dilemma. Either experiences themselves have empirical content or they do not. If they do, then there is some explanation of how this is so; and what is to prevent this same explanation being applied directly as an account of how beliefs acquire their empirical content without any relations to conscious experiences at all? If, on the other hand, experiences do not have empirical content, then it is at best unclear how they might be supposed to endow the beliefs with which they bear certain relations with any either, whatever these relations may be. The appearance of 2 McDowell, Mind and World, pp. 15ff. 205

Bill Brewer a dilemma here is illusory. For I shall argue that the correct account of the empirical content of experiences is such that its application to contents of a given kind is sufficient for these to be experiential contents themselves. Thus, although the second horn of the proposed dilemma is indeed unacceptable, its first horn is perfectly compatible with my premise that the empirical contents of beliefs depend upon their relations with experiences. The explanation of how it is that experiences themselves have empirical content is inapplicable directly to anything else. So the crucial role of experiences cannot be finessed by attempting to apply the same account in the domain of non-experiential beliefs. This reply to the dilemma objection would be completely vindicated by the following claim. A correct account of the content of perceptual experiences, in particular, of their reference to mindindependent spatial particulars, is both necessary and sufficient for explaining their status as conscious experiences? Although I am confident that this claim is true, I settle here for a slightly less ambitious justification for my response to the proposed dilemma. I shall argue that perceptual reference to particular mind-independent things is essentially experiential. So, again, nothing non-experiential could have its empirical content in just the way in which perceptual-experiential contents do.4 I assume that perceptual reference to determinate spatial particulars is possible; and that a person often knows that he is referring in his belief to a particular mind-independent thing when he is. Indeed, he is normally at least in a position to know this. For, suppose that he believes on the basis of perception that a is F. That is to say, asking himself how things stand in the world around him, he arrives at the judgement that a is F. Then, given the concept of belief, he can simply prefix this judgement with the operator 'I believe that', to arrive at the knowledge that he believes that a is FJ From here, he can knowledgeably infer that he is referring to a. 3 See N. Eilan, 'Self-Consciousness and Experience' (D.Phil, Thesis, Oxford University, 1988), for a sustained discussion of this kind of claim. 4 The argument which I am about to give is my development of P. F. Strawson's famous discussion of these matters in his Individuals (London: Methuen, 1959), chapter 1, part I. I draw heavily upon N. Eilan, 'SelfConsciousness', chapter 5, in which she also acknowledges the importance of conversations with Adrian Moore. 5 See G. Evans, The Varieties of Reference (Oxford: Clarendon Press, 1982), chapter 7; R. Gordon, 'Simulation Without Introspection or Inference from Me to You', in Mental Simulation, ed. M. Davies and T. Stone (Oxford: Blackwell, 1995), p. 60; and J. Roessler, 'Self-Knowledge and Belief (D.Phil. Thesis, Oxford University, 1996), chapter 2, for discussion of this so-called 'ascent routine', and its status as a source of knowledge. 206

Experience and Reason in Perception So, consider a person, S, with a perceptually-based belief about a particular mind-independent thing, a; and suppose that 5 actually knows that he is referring to a. Given what has just been said, stipulating this situation is legitimate. Now, assume, for reductio, that 5's Idea of a is purely descriptive.6 That is, his conception of which thing a is is exhausted by a wholly general description, 'The F, which purports to identify a by reference to its own properties and its spatial and other relations with other things which are also identified purely descriptively. Thus, his entertaining the Idea is quite independent of any experience of the object in question. Now, however detailed and extensive this description of a~may be, it is bound to be an epistemic possibility for 5 that F' is multiply satisfied, in the following sense. It is logically consistent with all that S knows that 'F' is satisfied by more than one thing. For he cannot knowledgeably rule out the possibility of a massive qualitative reduplication elsewhere in the universe of the relevant sector of his environment.7 So there is a possible world in which 'F° is multiply satisfied and everything which S actually knows is true. Thus, there is a possible world in which F' is multiply satisfied and S refers as he does to a; for that he does so is something which he knows. This is a contradiction. For in that case, 'The F' fails to refer. Hence 5's Idea of a cannot be purely descriptive after all.8 It must involve some kind of demonstrative component, with respect to which his experience is essential to his grasp of which object is in question.l) 6 See Evans, The Varieties, pp. 104ff., and P. Geach, Mental Acts (Bristol: Thoemmes Press, 1992), pp. 53ff., for introduction and elucidation of this notion of a person's Idea of an object. 7 The underlying equivalence here can be demonstrated as follows. Suppose that S is a person, p a proposition, and Ks the conjunction of every proposition which »S knows. It is epistemically possible for 51 that p O it is logically possible that [Ks & p] <=> {Ks, p} is consistent <=> 'Ks, therefore not-p' is invalid <=> S cannot argue validly from what he actually knows to not-p <=> S is not in a position knowledgeably to rule out that p. 8 It may be objected at this point that the epistemic possibility of multiple satisfaction can be ruled out if »S's descriptive Idea is of the following form: 'The unique G'. This is correct; but, in that case, reference failure due to the emptiness of this description is an epistemic possibility. For, again, S cannot knowledgeably rule out the possibility of massive reduplication with respect to G. So an appropriately adjusted version of my argument goes through. Note, also, that the argument does nothing to undermine the possibility of thought about spatial particulars by what might be called impure description, in which reference is secured, in part, by an embedded demonstrative, as in, for example, 'The red ball under that table'. '' This step in the argument requires further defence too. For it might be thought that either names or certain descriptions embedding indexicals 207

Bill Brewer Perceptual reference is therefore essentially experiential. Nothing non-experiential can refer to spatial particulars in the way in which experiences do; and, indeed, beliefs about spatial particulars of any kind have their contents only in virtue of their relations with essentially experiential demonstrative contents. My reply to the dilemma objection is therefore correct; and the first premise of my argument for the thesis that perceptual experiences must provide reasons for empirical beliefs is established. Certain beliefs about the spatial world have the contents which they do — that a particular mind-independent thing is determinately thus and so - only in virtue of their standing in certain relations with various perceptual experiences. Experience and Reason The second premise of my argument is that these crucial relations must be reason-giving relations, in the sense I outlined earlier: reason-giving from the subject's point of view, rather than from the provide an equally acceptable alternative to pure descriptions, in a way which finesses my appeal to essentially experiential perceptual demonstratives. What is wrong, for example, with the following Ideas: 'Frege'; and 'The ball in front of me'? I cannot respond in detail to these points here. My claim in each case, though, is that, insofar as the Ideas in question are immune to my objection from the possibility of massive reduplication, a complete account of what is involved in understanding them cannot avoid reference to perceptual experience. In the case of names, this is either a result of the connection between the subject's understanding of the name and his possession of a recognitional capacity for its bearer (if he is a Producer, in Evans' sense), or of the connection between his understanding of the name and the various experiences involved in his being informed about its bearer in testimony of some kind (if he is a Consumer). See Evans, The Varieties, chapter 11, for this distinction between Producers and Consumers in a name-using practice, and an outline of the account of proper names upon which my reply here draws. In the second case, experience comes in in the crucial role of demonstrative reference to physical things other than himself both in the subject's grasp of the spatial concepts figuring in such descriptive-indexical Ideas, and in his grasp of the first person pronoun itself. These issues require extended discussion though. A further source of objection might be the claim that experience is essential to demonstrative reference. After all, a blindsighted patient apparently refers to a particular object in his blind field when encouraged to point to it. Yet he has no experience of it. Here I simply insist that he does not understand any demonstrative thought he may appear to others to be having. For he does not know which object is in question. See L. Weiskrantz, Blind Sight: A Case Study and Implications (Oxford: Clarendon Press, 1986) for a comprehensive study of such cases. 208

Experience and Reason in Perception perspective of some external theorist. Why do I claim that the content-determining role of perceptual experiences necessarily involves their providing reasons in this sense for empirical beliefs? Well, suppose that the relevant content-determining relations between experiences and beliefs are not reason-giving relations; and consider a person, S, who believes that/), where this is supposed to be an empirical belief, about how things are in the mind-independent world around him. Since their relations with certain perceptual experiences play an essential role in the determination of the contents of empirical beliefs, there is a range of alternative such beliefs - beliefs which he might have had instead - whose difference in content with his actual belief that p would have been due entirely to their standing in the relevant relations with different perceptual experiences. Suppose that the belief that q is one of these. So, the situation is this. S actually believes that p, because his actual perceptual experiences determine this, as opposed to q, as the empirical content of his belief. He does not believe that q. Had his perceptual experiences been appropriately different, though, his position would have been precisely the reverse: he would have believed that q, and not believed that p. Yet the relevant contentdetermining relations between experiences and beliefs are not reason-giving relations. So S's actual perceptual experiences give him no more reason to believe that p than to believe that q. Thus, he has, and could have, no reason whatsoever to believe that p rather than that q, or vice versa. For, recall, nothing other than their relations with experiences decides between the two contents — this is how q was introduced. Which belief he actually has is due entirely to the course of his perceptual experience. Any supposed difference between believing that p and believing that q is therefore nothing to him; for there could be no reason for him to decide between them. So he does not really understand them as alternatives. Believing that p and believing that q are identical for him. Hence the supposedly content-determining role of S's perceptual experiences is empty. For there is nothing more, or less, to the content of a belief than the way the subject takes the world to be. Thus, if the relevant relations between experiences and beliefs are not reason-giving relations, then they contribute nothing to the determination of specific worldly truth-conditions for empirical beliefs. This all sounds a little abstract; but the crucial point is really quite simple. It can be brought home by the failure of a putative counterexample. Of course, it might be admitted, empirical beliefs draw essentially for their contents upon certain relations with perceptual experiences; but these need not be reason-giving relations, in the relevant sense. Perhaps experiences indicate worldly phe209

Bill Brewer nomena in virtue of various systematic causal relations in which they stand with them; and this empirical significance is transmitted in turn to the beliefs to which these experiences themselves give rise. Thus, empirical contents are secured for beliefs by their nowreason-giving relations with experiences. For, although these experiences stand in various relations with the worldly phenomena which they are thereby supposed to indicate, they do nothing to make beliefs about just those phenomena any more appropriate from the subject's point of view, than beliefs about any alternative such phenomena, which might reliably have caused the relevant experiences instead. A familiar, although in my view mistaken, account of our experiences of, and beliefs about, secondary qualities provides an illustrative example of the proposal. On this view, certain immediately recognizable sensations constitute our experiences of the secondary qualities of things around us, in virtue of the systematic causal relations between the two. The sensations in question are specific modifications of consciousness, the subject's recognition of which on any particular occasion, as tokens of the relevant types, is supposed to be unproblematic: subjectively evident, or 'given'. The corresponding secondary qualities of objects in the world are those microphysical properties, or massive disjunctions of such, which causally explain the occurrence of sensations of the type in question in normal observers under normal circumstances. Peacocke's introduction of 'primed predicates' helps to clarify the position.10 Red' experiences are those which are normally produced by red objects. Correlatively, red objects are those which normally produce red' experiences. Although the former claim serves to introduce the notion of observational predicate priming, it is the latter which captures the correct order of explanation. For token red' experiences are unproblematically identifiable as experiences of the same subjective type by their subject. Red objects are those which have the (perhaps massively disjunctive) microphysical property which normally produces red' experiences: this defines what redness in the world actually is. Red' experiences indicate the presence of red objects in virtue of the reliable relations between them. Beliefs that things are red acquire their content, in turn, as beliefs that things have just that colour - that microphysical property - on the basis of their relations with red' experiences. They are precisely the beliefs which are normally formed in response to such experiences. This is what makes them beliefs that the things in question are red. On this account, the crucial content-determining relations 10 C. Peacocke, Sense and Content (Oxford: Clarendon Press, 1983), pp. 20ff. 210

Experience and Reason in Perception between experiences and beliefs are not reason-giving relations. For having a red' experience in itself gives the subject no more reason to believe that there is something in front of him with the microphysical structure constitutive of something's being red than to believe that there is something there with any other such structure. It is simply the occurrence of a sensation of a particular identifiable type, which is intrinsically no more appropriately associated with that structure than with any other. Nevertheless, it is supposed to indicate the presence of just that microphysical property, as opposed to any other, because that is the property which happens to be its normal cause. Hence it is that property which he believes is instantiated when he believes that there is something red in front of him. His beliefs about redness therefore acquire their empirical significance in virtue of the relations they bear to red-type experiences, even though these experiences give him no reason to take the world to contain just that property rather than any other. Indeed, had it been a quite different property in the world which happened to be the normal cause of red' sensations, then his beliefs about redness would, on this account, have been beliefs to the effect that that is instantiated instead, regardless of the fact that what is supposed to provide such beliefs with their determinate content - namely their relations with red' experiences — is exactly the same in both cases. The consequence of this proposal, then, is that the putative source of the empirical content of beliefs about redness in the world is, as far as the subject himself is concerned, entirely neutral on which property of things their redness actually turns out to be. In believing that there is something red in front of him, a person is bound to be believing that there is something which is some way or other which things can be, and sometimes are, out there. Furthermore, given that redness is defined as the normal cause of red' sensations, he will normally be right that there is something that way in front of him. Yet he has not the slightest idea which way this is. Hence it must be wrong to claim that he nevertheless believes that the thing in question is just that way rather than any other. So the account under consideration fails after all to provide a satisfactory explanation of the contents of a person's beliefs about the colours of things around him. Two strategies are possible in reply to this switching argument.." First, it might be argued that the non-reason-giving account is not committed to a person's ignorance of the semantic values of certain of the concepts figuring in the contents of his empirical beliefs in " See C. Peacocke, 'The Limits of Intelligibility: A Post-Verificationist Proposal', Philosophical Review 97, (1988), for characterization and discussion of such arguments. 211

Bill Brewer the way I suggest. Second, it might be argued that, insofar as a nonreason-giving account is committed to this ignorance, it is nevertheless perfectly compatible with the relevant beliefs having just those contents all the same. I cannot possibly respond to everything which might be said here. I content myself with a few remarks about each of two possible versions of the first strategy, and a fairly obvious version of the second. Knowledge by Description. A first response to the switching argument is to deny that the present account necessarily consigns a person who believes that there is something red in front of him to ignorance of which property he thereby believes the thing to have on the following grounds. He may know perfectly well what it is for something to be red, say, by thinking of redness in the world (correctly according to the account in question) as, something like, 'the microphysical property which is causally explanatory of this type of sensation in normal observers under normal circumstances'. There are at least the following three difficulties with this line of response, though. First, there is an obvious danger of circularity in the proposed identification of redness as the cause of red' sensations in normal observers under normal circumstances. For the following two supplementary clauses have to be added in order to make sense of the descriptive identification as it stands: (i) normal observers are, inter alia, those in whom red things cause red' sensations in normal circumstances; (ii) normal circumstances are those in which red things cause red' sensations in normal observers. Yet without an independent constraint upon normality, the resultant identification as a whole is clearly useless. Second, it is implausible to suggest that a person's beliefs about the perceived colours of things around him essentially require this conceptual sophistication. For example, it is surely possible for a person without any explicit understanding of the idea of a normal observer under normal circumstances to believe that a certain item in his environment is red, and know perfectly well how he takes the thing to be. Third, even if the red things in the world are all and only those which produce red' experiences in normal observers under normal circumstances (or are disposed to do so), this is not the only, and certainly not the most basic, way in which a person thinks of their colour when he does so on the basis of such experiences. It is inappropriate to assign to all of a person's beliefs about the colours of things in the world around him the structural complexity which is currently being proposed. The most simple perceptually based beliefs about such matters are more naturally articulated by way of 212

Experience and Reason in Perception a demonstrative, 'that colour', than any causal description. Indeed, the inappropriateness of the present suggestion can be brought out by the fact that a person who knows precisely what colour he believes something is when he sees it in front of him in ordinary viewing conditions can perfectly rationally doubt whether the object in question has the microphysical property causally explanatory of certain specific appearances in normal observers under normal circumstances. In other words, any such description fails to capture his conception of which colour property is in question. So the way in which relations with colour experiences provide empirical significance for a person's beliefs about colours in the world is not by furnishing sensational effects by reference to which these colours can be identified as their normal causes. Conceptual Redeployment. A second version of the first strategy, of denying ignorance of semantic value on the non-reason-giving account, is to insist that a person's knowledge of which property he believes that the relevant object has, when he believes that there is something red in front of him, has already been provided for on the familiar account of secondary qualities under consideration, independently of any descriptive knowledge which the subject may have about the relation between his red' experiences and red things in the world. Indeed, it may be argued, it has been provided for in a way which has the advantage of explaining its notoriously problematic first person authority. For the subject can simply redeploy the concept of redness which figures in his first order belief about the world in the thought 'I am thinking that there is something red in front of me."2 Then, whichever property it is that his first order belief ascribes, which depends upon which microphysical property happens to be the normal cause of red' sensations and thereby constitute mind-independent redness, his second order belief will successfully self-ascribe a belief to the effect that the object in question has just that property. So he does, after all, know exactly how he believes the object to be, or at least he can authoritatively come to do so if only he turns his mind to the matter. This routine, then, provides an authoritative source of knowledge of the semantic values of his secondary quality concepts, which is consistent with the current account, on which these are determined by the non-reason-giving relations between secondary quality experiences and the beliefs in whose contents such concepts occur. So the switching argument is invalid. 12 T. Burge, 'Our Entitlement to Self-Knowledge', Proceedings of the Aristotelian Society 96 (1996); and C. Peacocke, 'Entitlement, SelfKnowledge and Conceptual Redeployment', Proceedings of the Aristotelian Society 96 (1996). 213

Bill Brewer I find this response quite unconvincing. Unless he already knows what redness is, a person cannot inform himself simply by reusing the word in an attempt to tell himself what it is. I am absolutely not insisting here that concept possession is in every case a matter of having some explicit, reductive definition of the concept in question in terms of more basic concepts. That would obviously be viciously regressive. It is just that a person does not grasp a concept unless he knows its semantic value, that is, the contribution it makes to determining the truth-conditions of beliefs in whose contents it occurs. Only then is he in a position genuinely to understand such beliefs; for only then can he appreciate what their truth-conditions must be. If he does not grasp this semantic value, then he cannot hope to help matters simply by taking on further beliefs, this time about his own beliefs, which invoke the very same concept, and which he therefore equally fails to understand. Although this charade guarantees that the subject's (pseudo-) self-ascription of a (pseudo-) belief about redness will be (pseudo-) true, it leaves him hopelessly ignorant about which truth this is supposed to be, which belief he has thereby self-ascribed. So the appearance that he thereby knows how he supposedly believes the world to be in the first place is just an illusion. Rather, the fatal ignorance at that first-order level is simply recycled at the level of the self-ascription itself. Once again, it is clear that we do not really have an account of genuine belief here at all, that is, belief with understanding, about the mindindependent world, about the subject's own beliefs, or about anything else. Natural Kinds. The second strategy in response to my switching argument is to accept that it does follow from the proposed account that a person is in a sense ignorant of certain of the details of the semantic values of the secondary quality concepts which he applies to the things around him; but to argue that this is perfectly compatible with these concepts figuring in the contents of his empirical beliefs all the same, which have precisely these contents in virtue of their non-reason-giving relations with his perceptual experiences according to just this account. Indeed, it might be insisted that ignorance of semantic value in this sense is a familiar phenomenon, perfectly compatible with grasp of a determinate empirical content. Consider, for example, beliefs about natural kinds. My beliefs about gold, say, are about just that chemical stuff, regardless of whether I know which stuff this is, and regardless of whether I can distinguish it in any way from the stuff which presents the same superficial appearance on twin-earth, or from fool's gold around here. 214

Experience and Reason in Perception This is a large and complex area, about which I cannot say anything entirely comprehensive or conclusive here. The first thing to do, though, is to distinguish this response to my criticism of the familiar secondary quality model from the first reply I considered above, on which knowledge of semantic value is supposed to be provided by description in these cases. The present suggestion is not that natural kind beliefs involve implicit descriptions of the form 'the microphysical structure of S)( S2)... Sn' (where the Si's are various samples of gold, say, with which the subject has had suitable contact — through perception, testimony or whatever) and that all beliefs about particular mind-independent things to the effect that they are determinately thus and so are to be assimilated to these descriptive natural kind beliefs. It is rather that beliefs about natural kinds constitute a ubiquitous example of cases in which a person succeeds in having beliefs to the effect that certain things have a particular microphysical structure in the absence of any detailed knowledge of which structure this is. Indeed, the present respondent will continue, everything involved in the subject's conception of which property is in question might be shared by a twinearth counterpart, who equally succeeds in referring to a determinate property in the beliefs he expresses using the word 'gold', whilst these are nevertheless beliefs to the effect that the relevant objects have a quite different chemical composition. Thus, a person might believe that certain things are gold - Au - as opposed to believing that they are twin-gold - ABC, say - even though, had it been precisely the reverse, he would himself have been none the wiser. Here there is determinacy in content which outstrips anything in the subject's knowledge of the appropriate semantic value. Yet these are clearly his beliefs. Hence this type of ignorance of semantic value is consistent with the understanding required of a person for him to be the subject of the beliefs in question. Similarly, it must be a mistake for me to infer, in the reductio argument given above, from a structurally similar position in the envisaged case of the secondary qualities, that the subject has no real understanding of his own beliefs. Genuine understanding is, after all, perfectly compatible with wholly non-reason-giving contentdetermining relations between experiences and beliefs. So it has not yet been established that perceptual experiences must provide reasons for empirical beliefs. I have a bold and a cautious reply back to this third line of response to the switching argument. The bold line is to stand by the original argument, and to conclude, therefore, that the account offered above of beliefs about natural kinds must be mistaken for the very same reason: it assigns determinate truth-conditions in a 215

Bill Brewer way which goes beyond the subject's own knowledge of the semantic values which purportedly determine them. A person's understanding of his own natural kind beliefs requires that 'gold' has different semantic values on two speaker's lips only if they have different conceptions of what gold is, in some sense. There are then a number of ways of developing the position, depending at least upon the answers given to the following two questions. (1) Do these conceptions necessarily supervene upon the physical condition of a person from the skin in? (2) Are differences in conception necessarily something about which the subject is infallible? I cannot possibly go into all the issues which are raised here;13 but the way in which I would be inclined to develop this bold reply is as follows.14 First, familiar arguments, paradigmatically from the causalexplanatory efficacy of the mental, for an affirmative answer to (1) are unconvincing. Second, familiar arguments, paradigmatically from the transparency of the subjective, for an affirmative answer to (2) are also unconvincing. Third, stories can be told on which Putnam's original insight that 'gold' (in his case 'water') has different semantic values on an earthling's lips and on those of his twinearth counterpart is compelling, even when the two counterparts are physically identical from the skin in, and neither can infallibly distinguish his position from that of the other.15 Fourth, they nevertheless have different conceptions of what the stuff is which they call 'gold'; and these conceptions contribute to their respective knowledge of the (different) truth-conditions of the beliefs which they each express using that word. Of course, this knowledge cannot require possession of the concepts of proton, neutron and electron, or any grasp of the way in which they are employed in the systematic construction of the periodic table in which Au has its characteristic place. For the word 'gold' in English had its current semantic value before these theoretical advances were made by anyone; and presently has it on the lips of competent English speakers without this knowledge of basic chemistry. The conception itself is 13 See Thought and Object, ed. A. Woodfield, (Oxford: Clarendon Press, 1982); and Subject, Thought and Context ed. P. Pettit and J. McDowell (Oxford: Clarendon Press, 1986), for a good orientation. 14 I take it that this is one way of filling out the suggestion in section 3 of the introduction to Subject, Thought and Context, ed. Pettit and McDowell, of a more radical response than Putnam's own 'composite' account to the original discussion of these matters, in H. Putnam, 'The Meaning of "Meaning"', in Mind, Language and Reality (Cambridge: Cambridge University Press, 1975). My sketch of this development owes a great deal to their treatment, and to other work by McDowell too. 15 Putnam, 'The Meaning of "Meaning".' 216

Experience and Reason in Perception likely to be essentially demonstrative, that stuff, or that metal, say: a legitimate conception of which stuff is correctly called 'gold', nevertheless, which differs, as required, between earthlings and their twin-earth counterparts. This is a perfectly consistent position. Its key component is a strongly externalist account of the nature of thinkers' conceptions of semantic values: these features of a person's mental life are themselves essentially world-involving demonstrative-recognitional capacities."' So the case of beliefs about natural kinds does not constitute one in which determinacy of belief content is really compatible with ignorance of the semantic values determining the associated truth-conditions. It is not one in which understanding transcends the subject's knowledge of these truthconditions at all. Thus, it does not constitute a counterexample to the original switching argument. The cautious reply admits the possibility of cases of beliefs about natural kinds which are as described in this third response to the switching argument. A person might have the understanding required for a specific empirical content to be the content of his belief, even if he is ignorant of certain of the details of the semantic values of the component concepts which determine these very truth-conditions, in the sense that he would have been none the wiser had these semantic values been different in some way. The key claim here, though, is that ignorance of semantic value, in this sense, is compatible with determinacy of content in such cases, only if possession of the relevant natural kind-like concepts depends upon possession of more basic observational concepts, grasp of which is incompatible with such ignorance of their semantic values. Thus, the model of beliefs about natural kinds exploited by this third line of response to the switching argument cannot possibly be universally appropriate. Hence, given the present dialectical situation, a non-reason-giving account of the content-determining role of perceptual experiences with respect to empirical beliefs cannot be correct in all cases either. There must be a basic class of observational concepts, for which a reason-giving account of the determination of their semantic values is imperative. I acknowledge that my discussion of both of the strategies which I distinguished above for responding to the switching argument is "' Evans, The Varieties, chapter 8. Of course, the precise nature of such capacities is a very difficult matter. Their possession presumably requires some sensitivity to the fact that any putative re-identification of the kind in question is defeasible by scientific investigation; and, relatedly, the ability to keep track of samples over certain changes in appearance, along, perhaps, with a rough idea of what types of change are compatible with continued instantiation of the kind. 217

Bill Brewer incomplete and unsatisfactory as it stands; and there are no doubt further variants of each to consider. Nevertheless, if what I say is at all on the right lines, and other variants can also be blocked, then any content-determining relations between perceptual experiences and the most basic empirical beliefs, at least, are necessarily reasongiving relations. The non-reason-giving alternative which I have been considering is unacceptable in such cases. Conjoining this with my first premise, that these relations are indispensable to genuine beliefs about the mind-independent spatial world, it follows that perceptual experiences must provide reasons for empirical beliefs. II My question now is this. How exactly do perceptual experiences provide reasons for empirical beliefs? The discussion will have to be fairly brief and, no doubt, disappointingly sketchy; but my answer, in a sentence, is this. Perceptual experiences are essential to a person's grasp of certain demonstrative contents, whose reference to particular mind-independent objects and properties is achieved in such a way that his simply entertaining these contents gives him a prima facie reason to endorse them in belief. The relevant perceptual demonstrative contents are expressible, in the first instance, only as 'that thing is thus'. Given the relevant facts about the way things are in the world around the perceiver, the direction and focus of his attention, and so on, which contents these are is beyond his control; and his grasp of such contents depends upon his actually standing in certain perceptual-attentional relations with the particular things to which they refer. I want to press further this issue, though, of exactly what is involved in his understanding, that is to say, his actually being the subject who is entertaining, these perceptual demonstrative contents. What does his knowledge of their truth-conditions consist in? More specifically, to begin with, what is involved in his grasp of the embedded singular demonstrative Idea of the particular object in question? I argued above that this Idea cannot be purely descriptive. Reference to spatial particulars is, in the most basic cases, essentially experiential. So the questions to ask now is exactly what a person's perceptual experiences contribute to his Idea of a particular object on a given occasion. The answer, at least in certain central cases, and especially where vision is the prominent modality, is that perceptual experience displays the spatial location of the object in question. This is what experience contributes to determinate reference to spatial particulars in these cases. For there is a fundamental interde218

Experience and Reason in Perception pendence between making a numerical distinction between qualitatively identical spatial particulars and assigning them different locations at a given time: for every time t, every mind-independent object which exists at t has a location at t, it has just one such location, and no two numerically distinct things (of the same sort) have the same location at t. So a key to reference to spatial particulars is knowledge of their location, or knowledge of what would make true an identification of the object in question with that at a given location. This constitutes knowledge of which object is in question. It provides the subject with a genuine Idea of that particular thing, which is employed in the particular content in question, and in any other content entertaining which involves thinking of the same thing in the same way. As I have effectively already established, if this location is given only in terms of certain spatial relations to other objects identified purely by description, then the possibility of massive reduplication returns to undermine the purported uniqueness of location and hence of determinate reference to any particular occupant. What is required, therefore, for genuine reference to a particular mind-independent thing, is knowledge of what would make true an identification of that object with one whose location is given relative to the subject making the reference himself. This may be given indirectly, as it would be by appeal to a suitably demonstratively anchored, impure description, such as 'the red ball in the box under that table'. But this possibility obviously rests upon that of a person's more direct identification of a particular location, relative to his own - in this case, that exploited in his demonstrative reference to that table. In general, I claim, perceptual demonstrative contents of the appropriate kind are precisely those which display the locations of the relevant objects directly in just this way. Perceptual reference and egocentric location come together, at least in the central cases which I am currently considering, in which determinacy of reference is secured by exploiting the interrelation between the numerical identity and location of spatial particulars at a time. This is (at least part of) the essential contribution made by a person's perceptual experiences at a given time to his understanding of certain demonstrative contents making reference to particular mind-independent things around him at that time.17 Now, if such a perceptual demonstrative content is to refer to a 17 There is of course a debate at this point about precisely what the relations of priority are, if any, between identifications, of various kinds, of objects and their locations in this basic case, in which the things and places in question are displayed in experience. See Strawson, Individuals, chapter 1; D. Wiggins, 'The Individuation of Things and Places (I)', 219

Bill Brewer particular mind-independent thing, in this way, then the subject's Idea of its location must be an Idea of a location in a world of places and things which are independent of his actual experiences of them. For his Idea of its location contributes essentially to his Idea of its identity. He must therefore understand that the very same location might have been spatially related in a quite different way with him: that thing there might equally have been perceived from any number of different points of view.18 Yet how is this condition to be met, since I insist that, in order to avoid reference failure due to the pos- \ sibility of massive reduplication, the location in question must be j given relative to the subject's own} j There are two lines of response to this problem. First, it might j be denied that a person's recognition that the same thing, at the I same place, might have been perceived from a different point of ; view really is a necessary condition upon perceptual demonstra- • tive reference to mind-independent spatial particulars. Second, it might be argued that the apparent inconsistency between the ; claim that this condition must indeed be met and my insistence that determinate reference to particular places and things is ultimately subject-relative in some sense is only illusory. I take it for 18 What if the thing in question is a part of the subject's body? Things get very complicated in this case. First, bodily awareness is likely to be involved in a peculiar way. Although I am sympathetic to the idea that perceptual systems are normally both exteroceptive and proprioceptive, as elaborated by J. J. Gibson in his The Ecological Approach to Visual Perception (Boston: Houghton Mifflin, 1979), the object of exteroception is not normally itself an object of proprioception, as it is in this case. Second, and relatedly, a person's body-parts are normally experienced as his own. See B. Brewer, 'Bodily Awareness and the Self, in The Body and the Self, ed. J. Bermudez, A. Marcel, and N. Eilan (Cambridge, MA: MIT Press, 1995), for a discussion of these two points and their interrelations. Nevertheless, if the body-part in question is anaesthetized and experienced only as 'that hand', as opposed to 'my hand', then the condition in the text applies. At least, in cases in which it is the hand's displayed location which informs the subject's perceptual reference to that particular hand, then he must grasp that that location might have been displayed as differently located on some egocentric reference frame. Proceedings of the Aristotelian Society, Sup. Vol. 37 (1963); M. Woods, 'The Individuation of Things and Places (II)', Proceedings of the Aristotelian Society, Sup. Vol. 37 (1963); Evans, The Varieties, chapter 6; and J. Campbell, 'The Role of Physical Objects in Spatial Thinking', in Spatial Representation, ed. N. Eilan, R. McCarthy and B. Brewer (Oxford: Blackwell, 1993). I am confident, though, that what I have to say here is independent of the detailed differences of view on these matters. 220

Experience and Reason in Perception granted that a third possible suggestion, to resist the essentially perspectival nature of spatial reference - to both places and things — is simply ruled out by my discussion of Strawson's argument from massive reduplication earlier (pp. 206-208 above), and its obvious extension to locations above. I consider each of the two genuine responses in turn. If the singular Ideas involved in perceptual demonstrative contents refer to mind-independent things, then it certainly follows that these are Ideas of things whose locations are in fact independent of the subject's experiences of them. It does not follow, though, according to a proponent of the first line of response above, that any person entertaining such an Idea must therefore recognize the independence of the location of the relevant mind-independent thing from his actual experiences of it in any way. He need have no understanding whatsoever, then, that that thing there might equally have been perceived from any number of different points of view. The claim that he must, she will continue, can only be sustained by appeal to an absurd principle along the following lines. If a person refers to an object, o, which is F, on the basis of a perceptual demonstrative Idea, /, then in grasping / he necessarily recognizes that its object is F. This principle is indeed absurd, not least because it entails a person's omniscience about the objects of such thoughts; but it is not required by my argument. The crucial point is rather that it is only on the basis of his grasp of its location - displayed relative to him in experience — that the subject has an Idea of that object at all, in the cases under consideration. Unlike its other properties, its location is not something he can go on to wonder about, having already identified the relevant object in thought. For his Idea of its location contributes essentially to his identification of which object is in question. Hence, which object is in question - which object his Idea is an Idea of - is determined in part by his Idea of its location. Thus, if he has no understanding whatsoever of the independence of this location from his actual experiential point of view upon it, then it is wrong to claim that he has an Idea of a mind-independent spatial thing at all. This is not the sort of thing which could be determined as the object of his Idea in that case. So the condition is quite genuine. A person's Idea of the relevant location must be an Idea of a location which might have been spatially related in a quite different way to him, in the following sense. He must actually grasp that that thing there might equally have been perceived from any number of different points of view. The first line of response is therefore untenable. Can anything really be made of the second suggestion either: to 221

Bill Brewer argue that meeting this condition is after all consistent with the requirement that determinate reference to particular locations is ultimately subject-relative in some sense? For if the location of a perceived particular is necessarily given in relation to the perceiver's own, then how can his perceptual demonstrative Idea involve any genuine understanding at all that that thing there, in just that place, might equally have been displayed as quite differently spatially related to him, from another point of view? To see how this might be done, it is crucial to get clear about a distinction due to John Campbell.19 There is both a monadic and a relational use of egocentric spatial terms such as 'to the left'/'to the right', 'above'/'below', 'in front'/'behind' and so on. Whenever a person uses these to give the location of something, its location is thereby specified in relation to something else: the terms are essentially relational. The subject can have more or less understanding of their relational nature, though. Representing sentences in which such terms occur as applications of the relation 'xRy', to begin with - 'x is to the right of y', as it might be - then the issue is over the appropriate range of the notional variable ly'. There is a primitive use of such sentences in which this is effectively fixed to the thinker himself, in his actual location at just that time, in such a way that he has no comprehension of the possibility of that very same 'relation' obtaining to anything else. So he is not really thinking relationally at all. His thought is more properly regimented 'R'a': 'a is to the right and a little in front', say, as opposed to 'a is to the right and a little in front of me'. For it is a minimal necessary condition upon discerning the genuinely relational structure in this thinking, 'aRi', where 'i' is a singular term referring to himself or his own present location, that he has some conception of what it would be for laRb', laRc' and so on to be true, for some appropriate range of alternative singular terms 'b', 'c', etc.20 The primitive use of egocentric terms, in which the subject has no such conception at all, is what Campbell calls 'monadic'; the associated concepts, or Ideas, of spatial 'relations' and particular locations are monadic spatial concepts, or Ideas. There are different types of genuinely relational uses of such terms, different types of relational spatial concepts and Ideas, which " J. Campbell, 'Spatial Decentring and Other Minds', forthcoming. I am also indebted at this point to Naomi Eilan for some very helpful suggestions. 20 This is of course an application to the current case of G. Evans' 'Generality Constraint'. See his The Varieties, esp. Chapter 4.3. The requirement is derived from Strawson, Individuals, esp. Chapter 3. 222

Experience and Reason in Perception vary according to the generality introduced by the range of the variable 'y': the range of appropriate alternative singular terms to 'i' — 'V, 'c', etc. above - which are such that the subject's grasp of the thought 'aRi' commits him to knowledge of what it would be for 'aRb', 'aRc' and so on to be true. Two significant types of relational egocentric thinking would be, first, that in which 'y' ranges over alternative possible points of view of the same thinking subject; and, second, that in which its range explicitly allows for generality across different thinkers. For present purposes, though, the basic distinction between monadic and relational egocentric spatial concepts, or Ideas, is enough. I have argued that determinate reference to particular locations is ultimately egocentric in some sense. That is to say, the embedded Idea of a particular place in the subject's environment must identify it in some way relative to himself, or his own present location. Our current problem is to explain how this requirement is consistent with his meeting the further condition upon perceptual demonstrative reference to particular mind-independent things, in cases where this exploits the subject's grasp of their location, that he should recognize that the very thing in question, just there, might equally have been perceived from any number of different points of view. Now, suppose that his egocentric perceptual identification of the location in question is purely monadic. So his conception of which place is in question is exhausted by its actual present spatial relations with him. It follows that he is incapable of any recognition that the thing in question, at just that location, might equally have been displayed as differently located — relative to him. For there is no degree of freedom in his thinking, along which to register the changes in his own position required to make sense of these alternative perspectives. He therefore fails genuinely to refer to any mind-independent spatial particular. His thought is not really thought about mind-independent reality at all. Put slightly differently, the difficulty here is that the perceptual demonstrative content in question is bound by a kind of idealism about space, on which there is nothing more to where things are than where they appear to be. Given that his knowledge of their location is supposed to contribute to his knowledge of which particular objects are in question, it follows that he can equally have no conception of these as existing independently of his particular experiences of them. His experience therefore fails altogether to display the way things are in a mindindependent world around him. Thus, if it is to refer to a particular mind-independent thing, a person's perceptual demonstrative, 'that thing - there - is thus', must comprise a singular Idea, 'that thing', which exploits a gen223

Bill Brewer uinely relational egocentric identification of the location of the thing in question. How exactly does this help? Well, a person whose perceptual demonstrative content is relational, in this sense, can immerse himself in his present perspective, so to speak, and entertain the corresponding purely monadic spatial content, 'that thing - therem - is thus'. This captures how things appear (spatially) to him - from that perspective, wholly immersed in it, and suspending any reflection upon it or its contribution to his experience. Yet in arriving at the appropriate monadic content on the basis of his prior understanding of the corresponding relational content, in this way, he grasps its grounding in the prior relational facts. In other words, this is to appreciate the joint dependence of how things currently appear to him (spatially) upon the way particular mind-independent things are actually distributed in the world around him, and his present location amongst them — his current perceptual point of view upon them. Equally, he is therefore at least in a position to simulate, to some extent, the monadic contents which would be associated with his taking up different points of view upon the same range of particular things, in the particular places they occupy. That is to say, he has the materials to construct, in imagination, the systematically varying monadic contents which he would arrive at by immersing himself in various alternative, possible but non-actual, perspectives upon the same mind-independent things, just where they are around him.21 In this way, I believe, he recognizes that the thing in question, at just that location, might equally have been displayed as differently located relative to him, from various different points of view. He is therefore able to meet this crucial condition upon genuine perceptual demonstrative reference to mind-independent things consistently with the requirement that the location exploited by his Idea of the particular thing in question should be specified egocentrically, in relation to his own. Parallel points apply, I believe, in cases where determinate reference is secured other than by appeal to perception of spatial location, in explanation of how essentially experi21 I do not mean to claim here that the subject must explicitly operate with a theory about the way in which spatial appearances vary with changes in his point of view. Rather, he has the potential, at least, to trip to and fro between a fixed relational conception of where a certain thing is relative to him, and both his actual, immersed monadic impression of its location, and various non-actual possible alternatives to it had he perceived that thing from a different point of view, simulated in imagination. Although his skill in this regard need not be at all well developed, this must at least be possible for him, in sense that his relational egocentric spatial Ideas are the essential ground for whatever such imaginative routines are actually engaged. 224

Experience and Reason in Perception ential reference can nevertheless be reference to mind-independent particulars.22 Something very similar also sustains the mind-independent significance of the predicative component of perceptual demonstrative contents. Experiential demonstratives, like 'that thing is thus', refer to mind-independent particulars - as we have just seen - of which they predicate, not just subjective appearances, but mind-independent properties, which are the categorical grounds of the relevant objects' powers to produce such appearances in appropriately placed perceivers. A given such content presents a particular thing as mind-independently, categorically thus. A person whose experience enables him to grasp this content can immerse himself in his own perspective, so to speak, and entertain the corresponding appearance - from there and in those circumstances - that that thing appears thus. In arriving at the appropriate appearance on the basis of his prior understanding of the corresponding perceptual demonstrative presentation of the way things mind-independently are, in this way, he grasps its grounding in the prior categorical facts. In other words, this is to appreciate the joint dependence of how things currently appear to him upon the way particular mindindependent things actually are, and his current point of view upon them and other relevant circumstances. Equally, he is therefore at least in a position to simulate the appearances, in this immersed sense parallel to monadic spatial contents, associated with his taking up different points of view on the same things, 22 The common structure here is this. Suppose that the contribution of experience to securing determinate reference to mind-independent particulars in the cases under consideration is its displaying the relevant object's characteristic (f>. In the central cases I have been discussing, <f> is spatial location; equally, it might be timbre, for example, in auditory reference to a particular member of an unseen wind quintet. My development of Strawson's argument in section I implies, on the one hand, that this contribution is to make available to the perceiver an essentially experiential demonstrative Idea of the object in question, such as 'that (f>-thing'. The mind-independent reference of this Idea, on the other hand, requires that his grasp of the characteristic <p in question should not be exhausted by the way in which it is actually presented in his current experience. The apparent tension between the two is to be resolved by the subject's grasp of the actual experiential appearance of <f), from immersed within his present perspective and suspending any reflection upon it or its contribution to his experience, as the joint upshot of the mind-independent <f> of the object in question and the relevant features of his particular perceptual perspective upon it. As will become clear, it is precisely this understanding which I think constitutes his prima facie reason to endorse the demonstrative content in question in belief. 225

Bill Brewer with the same properties, in different circumstances. That is to say, he has the materials to construct, in imagination, the systematically varying appearances of those things' being just the way they are from various alternative, possible but non-actual, perspectives and in various alternative, possible but non-actual, circumstances. Thus, he recognizes that the thing in question, being just as it is, might equally have appeared differently in different circumstances. This is part of what is involved in its being the case that his initial perceptual demonstrative content predicates a genuinely mind-independent, categorical property of the particular thing in question. With respect both to the singular and to the predicative components of perceptual demonstrative contents, then, the genuine mind-independence of their reference resides in the subject's recognition of what they present as the categorical ground of the corresponding immersed monadic contents and appearances from his present point of view in the present circumstances, and, equally, of the various alternative monadic contents and appearances associated with possible but non-actual points of view and circumstances, some of which he may be able to grasp in imaginative simulation. Hence, a subject of such contents necessarily recognizes that the way things currently appear to him is the joint upshot of the way things are anyway, in the mind-independent world around him, and his current point of view upon them and other relevant circumstances of perception. It is this, I claim, which provides his prima facie reason to endorse those very contents in belief. For, simply in virtue of entertaining perceptual demonstrative contents of this kind, he recognizes that it is that thing, there in relation to him, and mind-independently thus, which is currently displayed - from where he is and in those circumstances - as apparently thus monadically there. That is to say, he understands that his current apprehension that things are thus and so is in part due to the very fact that they are. He recognizes the relevant content as his apprehension of, or epistemic openness to, the facts. So, a person's experiences contribute essentially to his grasp of certain perceptual demonstrative contents. These contents refer to particular mind-independent things in the world around him, of which they predicate determinate mind-independent properties. In doing so, they give him a prima facie reason to endorse those very contents in belief. Simply in virtue of grasping the content that that thing - there - is thus, he has a prima facie reason to believe that that thing is indeed thus; for he necessarily recognizes that his entertaining that content is a response to that thing's actually being thus, given his location and present circumstances. 226

Experience and Reason in Perception I argued in section I that perceptual experiences must provide reasons for empirical beliefs; and in section II I have outlined how they do so. As I put the thesis right at the outset, the correct account of the sense in which perceptual experiences are experiences of mind-independent spatial things - that is to say, the correct account of the reference of perceptual demonstratives to mind-independent particulars and their properties - is itself an account of the way in which such experiences provide prima facie reasons for the perceiver's beliefs about the way things are in the world around him. 227

Intentionality as the Mark of the Mental TIM CRANE I Brentano's Thesis 'It is of the very nature of consciousness to be intentional' said Jean-Paul Sartre, 'and a consciousness that ceases to be a consciousness of something would ipso facto cease to exist.'1 Sartre here endorses the central doctrine of Husserl's phenomenology, itself inspired by a famous idea of Brentano's: that intentionality, the mind's 'direction upon its objects', is what is distinctive of mental phenomena. Brentano's originality does not lie in pointing out the existence of intentionality, or in inventing the terminology, which derives from scholastic discussions of concepts or intentiones.2 Rather, his originality consists in his claim that the concept of intentionality marks out the subject matter of psychology: the mental. His view was that intentionality 'is characteristic exclusively of mental phenomena. No physical phenomenon manifests anything like it.'3 This is Brentano's thesis that intentionality is the mark of the mental. Despite the centrality of the concept of intentionality in contemporary philosophy of mind, and despite the customary homage paid I am very grateful to Michael Martin and Gregory McCulloch for many conversations which have greatly influenced my views on this subject. I am also grateful to the participants in the discussion at the Royal Institute of Philosophy meeting in February 1997, and to Victor Caston, Katalin Farkas, Marcus Giaquinto, Paul Horwich, Michael Martin, Gregory McCulloch, Paul Noordhof and Scott Sturgeon for comments on earlier versions of the paper. 1 The Psychology of the Imagination (London: Routledge, 1995; originally published 1940) p. 211. 2 For the origins of the concept of intentionality, see Christian Knudsen, 'Intentions and Impositions', in The Cambridge History of Later Medieval Philosophy, ed. A. Kenny et al. (Cambridge: Cambridge University Press, 1982); Victor Caston, Aristotle on the Problem of Intentionality', forthcoming. For a general survey, and further bibliography, see Tim Crane, 'Intentionality', forthcoming in the Routledge Encyclopedia of Philosophy, ed. E. J. Craig (London: Routledge, 1998). 3 F. Brentano, Psychology from an Empirical Standpoint (London: Routledge, 1995; originally published 1874), p. 89. 229

Tim Crane to Brentano as the one who revived the terminology and placed the concept at the centre of philosophy, Brentano's thesis is widely rejected by contemporary philosophers of mind. What is more, its rejection is not something which is thought to require substantial philosophical argument. Rather, the falsity of the thesis is taken as a starting-point in many contemporary discussions of intentionality, something so obvious that it only needs to be stated to be recognized as true. Consider, for instance, these remarks from the opening pages of Searle's Intentionality: Some, not all, mental states and events have Intentionality. Beliefs, fears, hopes and desires are Intentional; but there are forms of nervousness, elation and undirected anxiety that are not Intentional ... My beliefs and desires must always be about something. But my nervousness and undirected anxiety need not in that way be about anything.4 \ Searle takes this as obvious, so obvious that it is not in need of further argument or elucidation. And many others agree with him.5 Brentano's thesis is normally rejected for one or both of the following reasons. First, it is supposed to be obvious that there are both intentional and non-intentional mental states: intentionality is not necessary for mentality. Non-intentional mental states can either be of the kind Searle mentions (emotions or moods, like undirected anxiety) or they are the so-called purely 'qualitative' mental phenomena - states which have 'qualia' of which sensations (like pains) are the most commonly cited examples. Both kinds of example are mentioned in an implicit rejection of Brentano's thesis by Louise Antony: while mental items like beliefs and desires clearly have objects or contents (an idea is an idea of something, and a desire is a desire for something), things like pleasures, pains, moods and emotions don't, on the face of it, appear to be about anything at all.6 Antony remarks in passing that the opposing view — in effect, that pains, moods and emotions are intentional - 'seems counter-intuitive'. The second reason for the rejection of the thesis is that there are now-mental phenomena which exhibit intentionality: intentionality 4 John R. Searle, Intentionality (Cambridge: Cambridge University Press, 1983), p. 1. 5 A representative example of this position is Colin McGinn, The Character of Mind (Oxford: Oxford University Press, 1982), chapter 1. 6 Louise Antony, 'What It's Like to Smell a Gardenia', The Times Literary Supplement 4897 (7 February 1997), p. 25. See also Fred Dretske, Naturalizing the Mind (Cambridge, MA: MIT Press, 1995), p. 28. 230

Intentionality as the Mark of the Mental is not sufficient for mentality. Examples are more controversial here, but we find phenomena such as the disposition of plants to move towards the source of light offered as primitive non-mental forms of intentionality.7 Not every philosopher who rejects Brentano's thesis rejects it for both of these reasons, but it is nonetheless fair to say that there is a tacit consensus that the thesis should be rejected. But this consensus raises a puzzling historical and exegetical question. If it is so obvious that Brentano's thesis is false, why did Brentano propose it? If a moment's reflection on one's states of mind refutes the thesis that all mental states are intentional, then why would anyone (including Brentano, Husserl, Sartre and their followers) think otherwise? Did Brentano have a radically different inner life from the inner lives of contemporary philosophers? Or was the originator of phenomenology spectacularly inattentive to phenomenological facts, rather as Freud is supposed to have been a bad analyst? Or - surely more plausibly - did Brentano mean something different by 'intentionality' than what many contemporary philosophers mean? The question of what Brentano and his followers meant by 'intentionality' is an important one, both for our understanding of the origin of current debates, and, relatedly, for our conception of these debates themselves.8 However, my concern in this paper is not with Brentano's theory of intentionality, but with a more general question: what would you have to believe about intentionality to believe that it is the mark of the mental? I argue here that if we think of intentionality in the light of this question, a conception of the mental begins to emerge which abandons some of the usual assumptions of contemporary philosophy of mind. The rest of this paper falls into three parts. In the next two parts I examine the standard counterexamples to Brentano's thesis - certain kinds of sensations and emotions. I argue that they are not genuine counterexamples, and I sketch a conception of intentionality which arises from the discussion of these examples. In the final part, I claim that intentionality, properly understood, should be thought of as exclusive to the mental domain, and I conclude with some more speculative remarks about the significance of the question: why do we need a mark of the mental at all? 7 For examples of this kind of approach, see Fred Dretske, Knowledge and the Flow of Information (Cambridge, MA: MIT Press, 1981); B.Enc, 'Intentional States of Mechanical Devices', Mind 91 (1982). 8 For a helpful attempt to relate Brentano's concerns to current debates, see Dermot Moran, 'Brentano's Thesis', Proceedings of the Aristotelian Society, Supplementary Volume 70 (1996). 231

Tim Crane II The Intentionality of Sensation Since 'intentionality' is a technical term, it is standard practice when introducing it to use some slightly less technical synonym or gloss. In contemporary philosophy this is often done by saying that intentionality is the about-ness or of-ness of mental states.9 Intentional states are those which are about or of things, normally things other than themselves. So one might demarcate the class of intentional states by considering a mental state and asking 'what is it about?' If the question doesn't make much sense, or if it has the obvious answer 'nothing', then the state is classified as non-intentional. Consider a pain you may have in your ankle; what is it of or about? Silly question: it isn't of or about anything. And so, since intentionality is of-ness or about-ness, pain is not intentional. In this way, we find Colin McGinn arguing that 'bodily sensations do not have an intentional object in the way perceptual experiences do' on the grounds that 'we distinguish between a visual experience and what it is an experience of; but we do not make this distinction in respect of pains'.10 This is one quick way to arrive at a denial of Brentano's thesis. It certainly sounds awkward to talk of a distinction between a pain and what a pain is 'of or 'about'. But all this means is that those who follow Brentano in holding intentionality to be the mark of the mental - call them 'intentionalists' - will not gloss the concept of intentionality solely in terms of 'of-ness' or 'about-ness'. Intentionalists must introduce what is involved in the phenomenon of intentionality in another way. How should they do this? Brentano's own view was that every mental phenomenon exhibits what he called 'intentional inexistence'. The term 'inexistence' has little or nothing to do with the fact that intentional states (or 'acts') can be about objects which do not exist.11 Rather, the term describes 9 See (for example) Martin Davies, 'Philosophy of Mind', in Philosophy: A Guide Through the Subject, ed. A. C. Grayling (Oxford: Oxford University Press, 1995); Searle, Intentionality, p. 1; Jerry Fodor, Psychosemantics (Cambridge, MA: MIT Press, 1987), chapter 4. 10 McGinn, Character of Mind, p. 8. 11 For examples of this common misunderstanding of Brentano, see Daniel C. Dennett, Content and Consciousness (London: Routledge and Kegan Paul, 1969), p. 21; Michael Tye, Ten Problems of Consciousness (Cambridge, MA: MIT Press, 1995) pp. 94-5. A useful account of Brentano's views is contained in chapter 1 of David Bell's Husserl (London: Routledge, 1990). For the scholastic views to which Brentano is alluding, see Knudsen, 'Intentions', and John Marenbon, Later Medieval Philosophy (1150-1350): An Introduction (London: Routledge 1987),