perceptual contact, or of sense-modally specific contact, is constituted by a further perceptual fact of an experientially richer or more determinate kind—a fact of which the fact of contact can be thought of as either an integral element or a determinable version. But even this does not affect the basic point. For, whether he pursues this option or not, the theorist is still obliged to say that, in some suitably specified form, the Φ-terminal perceptual relationship, is psychologically fundamental.
As I have indicated, this fundamentalist view of the Φ-terminal relationship is not something which, in the way I have defined it, SDR formally entails: formally, the theory only excludes those forms of constitution which yield psychological mediation. But the point is that, with the exception allowed for—an exception which is not in fact in conflict with the fundamentalist view—there are no other constitutional options that one could begin to take seriously. Once someone has excluded the form of constitution definitive of psychological mediation, he cannot, in practice, avoid recognizing the Φ-terminal relationship (suitably specified) as a fundamental aspect of the psychological situation. From now on, I shall simply assume that this fundamentalist view is an integral part of the SDR-position.
This said, there is one important respect in which the fundamentalist implications of SDR must not be exaggerated. In denying that Φ-terminal perceiving is psychologically mediated, the SDR-theorist has indeed to recognize it as psychologically fundamental—fundamental at the psychological level of description. But he does not have to take it to be fundamental absolutely. For his acceptance of SDR as the correct account of what perception is in its own psychological terms leaves him free to adopt a reductive account of the psychological facts themselves. It leaves him free to claim that the facts of perception are ultimately constituted by facts of a non-psychological kind; and it leaves him free to endorse a similar account for psychological facts quite generally. The main option here would be to adopt some form of strong and comprehensive materialism, which takes all psychological facts to be ultimately constituted by physical facts—for example, about the physiology and functioning of the human organism, and the interaction of the organism with the physical environment. This reductive approach to mentality is not, of course, the exclusive prerogative of SDR: it is equally available under BRT, and, indeed, certain forms of BRT-account may be thought positively to facilitate it. The only point in underlining its availability to the SDR-theorist is that this is something which his fundamentalist claim at the psychological level is liable to obscure.
In taking the Φ-terminal relationship to be psychologically fundamental,
end p.16
 
the SDR-theorist is not, then, excluding a reductive account of mentality itself, and this is something which it is crucial to bear in mind if we are to have a correct understanding of the nature of his position. At the same time, we should not expect this point to feature very prominently in our subsequent discussion. The issue between SDR and BRT is, in itself, to do with how things stand at the psychological level—with how the perceptual relationship is to be understood in its own psychological terms. It is inevitable that, for the most part, the considerations which bear on this issue will be ones which arise at that same level—considerations which do not relate to the question of the metaphysical status of the psychological realm itself. Even so, the availability of the reductive approach will be of relevance in one area of our discussion, as we shall shortly see.
II
There are two final preliminary points that I need to make, though not ones which are in any way connected.
The first concerns the implications of BRT. BRT, as I have said, takes the Φ-terminal perceptual relationship to break down into two components, one of which is the subject's being in a certain psychological state—a state which is not, in itself, perceptive of the relevant physical item—and the other of which comprises certain additional facts, but ones which do not involve anything further about the subject's psychological condition at the relevant time. Now, in requiring the relevant psychological state to be one which is not, in itself, perceptive of the relevant physical item—one which does not, on its own, suffice to put the subject into perceptual contact with that item—BRT is, in effect, requiring it to be one which is not, in itself, physically perceptive at all—to be a state which is logically capable of realization without there being anything physical perceived. This is not because there is any general difficulty in understanding how a state which is, in itself, physically perceptive could be mediationally involved in the perceiving of a physical item of which it is not in itself perceptive. For a state which is in itself perceptive of one physical item might be mediationally involved in the perceiving of another. But once we have reached the point of Φ-terminality—the point where there is no further physical item which is more immediately perceived—the only way in which the psychological state which is fundamentally involved in the perceiving of the relevant physical item could turn out to be in itself physically perceptive would be by being, in itself, perceptive of that item.
The second point concerns the framework of our discussion. I have
end p.17
 
already stressed that, for the time being, I am taking for granted the truth of physical realism. But, in turning to the issue between SDR and BRT, there is one other assumption which I shall provisionally make. For I shall provisionally assume that, whichever of these theories gives the right account of it, we do have genuine perceptual access to the physical world, and that, specifically, under this access, physical items become perceptible to us in just those kinds of circumstance, and by the use of those sense organs, that we ordinarily suppose—so that, for example, the relationship in which I now stand to the piece of paper on which I am writing is one of genuine seeing, and the relationship in which I now stand to the voices of the children in the garden is one of genuine hearing. Theoretically, this common-sense assumption could turn out to be false. For it is conceivable that the relationship in which we stand to the things which we suppose ourselves to perceive should fail to meet the requirements implicit in our actual concept of perception. But while this is a possibility which we may need to consider at some point, it will be convenient if, for the time being, we put it on one side, and work with the presumption that our perceptual capacities are what we ordinarily take them to be.
end p.18
 
Abstract: Strong direct realism (SDR) has the resources to meet the Causal Argument from hallucination, which I once thought decisive. But it fails for a different reason. When a physical item is Φ-terminally perceived, it sensibly appears to the subject in a certain way. Or put the other way round, the subject perceives the item in a certain phenomenal manner. I call this 'phenomenal manner of perceiving phenomenal content'. Like any other theory of perception, SDR has to be able to give an adequate account of what phenomenal content is and how it relates to the securing of perceptual contact with the relevant physical item. There are three options here for SDR. The first is what I call 'the presentational view'. This holds that the qualitative ingredients of phenomenal content are directly drawn from the physical items Φ-terminally perceived, so that when a physical item sensibly appears to a subject to possess a certain quality Q, the featuring of Q in the phenomenal content of the perception is the featuring of the very instance of Q that occurs in the physical item itself. The second option is what I call 'the internalist view'. This holds that, although the Φ-terminal perceptual relationship with the physical item is something psychologically fundamental, and phenomenal content is the manner in which this relationship holds, the qualitative ingredients of such content are ontologically separate from their physical counterparts, so that the featuring of a quality in such content is not the featuring of some physical instance of it. The third option is what I call the 'modified presentational view', which holds that the featuring of a quality in phenomenal content is sometimes to be construed in a presentational way, and sometimes in an internalist way. The presentational view fails because it does not accommodate cases of non-veridical perception, in which the sensible appearance of the perceived item is at variance with its actual character. The modified presentational view fails because of its hybrid character; for once the need for an internalist account is recognized for cases of non-veridical perception, there is irresistible pressure to extend the same treatment to veridical perception too. And the internalist view fails because it does not permit a coherent account of how perceptual contact and phenomenal content fit together. Since all three options fail, and since there is no other remotely plausible account available to it, SDR must be rejected.
Keywords: appearance,causal argument,direct realism,hallucination,internalism,perception,perceptual contact,phenomenal content,presentational,sensible appearance,veridical
Part Two An Examination of Strong Direct Realism
1 The Issue Before Us
I
For the time being, we are taking for granted the framework of physical realism—the view that the physical world is something whose existence is logically independent of the human mind, and something which is, in its basic character, metaphysically fundamental. Within this framework, the tradition offers two rival views of the nature of physical-item perception. Thus, on the one hand, there is the direct realist position, which claims that our perceptual access to the physical world is direct. And, on the other hand, there is the representative theory, which claims that our perceptual contact with physical items is always mediated by some form of mental representation. We saw, in Part One, that this traditional issue needs to be reshaped, since the claim of direct access is open to two crucially different interpretations, and, for each of the resulting versions of direct realism, there is, in appropriate opposition, a corresponding version of the representative theory. As a result, the traditional issue has turned into two issues: an issue between a weak (or modest) version of direct realism (WDR) and a narrow (or restrictive) version of the representative theory (NRT); and an issue between a strong (or full-blooded) version of direct realism (SDR) and a broad (or flexible) version of the representative theory (BRT). For reasons which I explained, it is the second of these issues—between SDR and BRT—that we need to consider first.
The conflict between these two positions is over the question of whether physical-item perceiving is subject to a certain form of decomposition. We have agreed to speak of a subject as Φ-terminally perceiving a physical item when he perceives that item and when there is no other physical item which (in relation to his perceiving of the first) he perceives more immediately.
end p.19
 
Then BRT claims that Φ-terminal perceiving is always psychologically mediated. This means that, for any subject S, physical item x, and time t, if S Φ-terminally perceives x at t, then there is a psychological state (type-state) Σ, which is not in itself x-perceptive, such that S's perceiving of x at t breaks down into (is constituted by the combination of) two components: one component consists in S's being in Σ at t; the other comprises certain additional facts, though ones which do not involve anything further about S's psychological condition at t. In contrast, SDR claims that Φ-terminal perceiving is always psychologically direct. This means that, for any subject S, physical item x, and time t, if S Φ-terminally perceives x at t, then S's perceiving of x at t does not break down into two components in this BRT-way: it is not constituted by the combination of S's being in this sort of psychological state and these sorts of additional fact. The reason why BRT counts as the broad (or flexible) version of direct realism is that, apart from requiring it to be not, in itself, perceptive of the relevant physical item, it does not impose any conditions on the nature of the mediationally relevant psychological state, whereas the narrow (or restrictive) version, NRT, requires this state to involve the occurrence of an internal object of awareness. And the reason why SDR counts as the strong (or full-blooded) version of direct realism is that it rules out psychological mediation of any kind, whereas the weak (or modest) version, WDR, only rules out that special kind of mediation postulated by NRT.
As we noted, BRT could equally be formulated as a claim about all cases of physical-item perceiving, not just those that are Φ-terminal, since any form of psychological mediation which applies at the point of Φ-terminality is bound to apply to the perceiving of any further physical item which the Φ-terminal perceiving perceptually mediates. And, as we also noted, in excluding the decomposition definitive of psychological mediation, an acceptance of SDR in effect involves taking the Φ-terminal perceptual relationship to be something psychologically fundamental; and, from now on, we are assuming this fundamentalist view to be an integral part of the SDR-position. Finally, we noted that, in requiring the psychological state which is mediationally involved in Φ-terminal perceiving to be one which is not, in itself, perceptive of the relevant physical item, BRT is in effect requiring it to be one which is not, in itself, physically perceptive at all—a state which can be realized without there being anything physical perceived.
Having identified the issue that immediately concerns us, we must now turn to the task of trying to resolve it. Of course, this resolution need not take the form of simply establishing the truth of one of the rival theories
end p.20
 
and refuting the other. For it might turn out that Φ-terminal perceiving comes in two quite different forms, and that one of these forms requires an account along the lines of SDR, and the other requires an account along the lines of BRT. We cannot assume in advance that the truth will be tidy.
II
Of the two rival theories, SDR is likely to strike us, initially, as the more attractive option. This is partly because it is the view of 'common sense'—the view which is implicit in what we take for granted prior to philosophical reflection. Thus we ordinarily think of our perceptual awareness as reaching out to its external targets in a wholly straightforward way—a way which would not leave room for any kind of decomposition at the psychological level. Indeed, as we shall see, our ordinary view of the situation commits us to accepting SDR in an especially strong form. But there is also another and more significant factor which puts SDR in a favourable light. For when we do philosophically reflect on the situation, our first response is to wonder how we can even make sense of the BRT-account. What BRT claims is that, when someone Φ-terminally perceives a physical item, his perceptual contact with it breaks down into his being in a certain psychological state and certain additional facts. The psychological state is one which is not, in itself, perceptive of the relevant physical item, and this means, in practice, one which is not, in itself, physically perceptive at all. At the same time, the additional facts, whatever they comprise, do not involve anything further about the subject's current psychological condition. But if the psychological state does not, in itself, secure perceptual contact with the physical world, and if the additional facts add nothing further to the subject's psychological condition, it is hard to see how the subject's awareness can be thought of as reaching beyond the boundaries of his own mind, and so hard to see how his relationship with the relevant physical item can be thought of as genuinely perceptual. Of course, this is just how things strike us initially, on a quick glance at the situation set out in broad terms: it could well turn out that the difficulty evaporates once we reflect on things in more depth and in greater detail. But what at least has to be acknowledged is that, in this initial perspective, SDR presents itself as the more promising approach.1
1 It has also, in recent years, gained considerable ground over its rival in philosophical esteem. Among those philosophers whose views we can see as implicitly endorsing it are Paul Snowdon and John McDowell (though, in the case of McDowell, it is not always easy to disentangle his views about the directness of perception from his views about the directness of perceptual knowledge). Thus see Paul Snowdon, 'Perception, vision, and causation', Proceedings of the Aristotelian Society, 81 (1980-1), 175-92, and John McDowell, 'Criteria, defeasibility, and knowledge', Proceedings of the British Academy, 68 (1982), 455-79. Both articles are reprinted in J. Dancy (ed.), Perceptual Knowledge (Oxford: Oxford University Press, 1988), in McDowell's case with certain revisions and additions.
end p.21
 
Despite its initial attractions, my own view is that SDR is mistaken, and this is the conclusion which I shall be trying to establish in the present phase of our discussion. The main focus of this phase will be on a certain aspect of the experiential content of perception; and the basic thrust of my argument will be that SDR cannot provide a satisfactory account of the nature of this aspect and of the role it plays, or the place it occupies, in the obtaining of the perceptual relationship. But before I turn to this, I want to consider an argument which attacks SDR from a quite different direction. In general conception, the argument coincides with one which has been strongly defended by Howard Robinson in a recent book, though the issue to which he explicitly relates it is not quite the same as the SDR-BRT conflict on which we are focusing here.2 I too deployed the argument in an earlier book, and explicitly in relation to the present issue, and saw it then as decisive.3 In the context of the present discussion, the pursuit of this topic will form something of a digression. But I regard the argument as sufficiently interesting in its own right to merit a detailed examination.
The argument will focus on the phenomenon of hallucination, and on how we should understand its distinction from genuine perception. A hallucinatory episode, in the sense that concerns us, is an experience which is subjectively like (introspectively indistinguishable from) an event of genuine (physical-item) perception, but in which no physical item is actually perceived. Obviously, the rival theories will see the distinction between perception and hallucination in quite different ways. BRT takes the psychological states which are fundamentally involved in perception to be ones which are not in themselves physically perceptive, and hence to be states which are logically capable of occurring in both perception and hallucination. So the BRT-theorist will say that, at least for one category of hallucinations, there is no difference between the psychological character of perceptual and hallucinatory events at the fundamental level of description.
2 Howard Robinson, Perception (London: Routledge, 1994), ch. VI. See also his 'The general form of the argument for Berkeleian idealism', sect. IV, in J. Foster and H. Robinson (eds.), Essays on Berkeley (Oxford: Oxford University Press, 1985). A primitive version of the argument is found in C. D. Broad, 'Some elementary reflexions on sense-perception', Philosophy, 27 (1952), 3-17, though without his wholehearted endorsement.
3 In my Ayer (London: Routledge & Kegan Paul, 1985), 147-9, 161.
end p.22
 
SDR, in contrast, holds that, at the point of Φ-terminality, the perceptual relationship is something psychologically fundamental. So the SDR-theorist will say that the psychological states which are fundamentally involved in perception are radically different from those involved in hallucination. He will say that such states are in themselves physically perceptive and are therefore logically incapable of realization in the context of hallucination. It is on this sharp contrast between the views of BRT and SDR that the argument turns.
2 The Causal Argument
I
Imagine that we have a minute radio-controlled device which can be attached to someone's optic nerves (perhaps, strategically, at the point of the optic chiasma), and which, thus located, can be used by an external operator to exercise full control over the neural signals which pass to the brain. This means that when the device is activated and the operator has assumed full control, the light which enters the subject's eyes has no influence on his central nervous system: the neural signals which pass to the brain are wholly determined by the functioning of the device, which in turn is fully controlled by the radio signals which the operator (no doubt with the assistance of a powerful computer) transmits to it. Now let us suppose that, without his knowledge, the device has been implanted in a particular subject, Henry, and that one afternoon, Henry is sitting on a river bank, with the controller discreetly watching events from a nearby cottage. Initially, the device is switched off, and, with his visual system working normally, Henry happens, at a certain moment, to catch sight of a salmon as it leaps out of the water. A little later, Henry's gaze is fixed on the spot where he saw the salmon leap, and the controller, judging this to be an opportune moment, switches on the device and programmes it to produce neural signals exactly like those which were photically produced by the encounter with the salmon, and its surroundings, on the earlier occasion. (How he manages to engineer this exact neural match need not concern us; if necessary, we could suppose that he succeeds, at least partly, by chance.) Let us also assume, what is perfectly plausible, that the neural signals on the two occasions have exactly the same effect on Henry's brain, so that, down to the last detail, the relevant brain state involved in the seeing of the salmon on the first occasion is reproduced on the second.
end p.23
 
Now, following through this thought-experiment, what should we expect to happen, psychologically, as a result of the controller's intervention? Well, given their exact similarity to what occurred earlier, it seems reasonable to assume that the artificially induced neural signals and the brain responses they elicit will produce a visual experience which subjectively matches the earlier perception, so that it will be with Henry, subjectively, exactly as if he sees a qualitatively identical salmon making a qualitatively identical leap. But, crucially, it also seems reasonable to suppose that, in addition to subjectively matching (to being introspectively indistinguishable), the two resulting mental episodes will be, at the fundamental level of description, of exactly the same psychological type—instances of exactly the same psychological state. For, granted that the relevant neural factors are qualitatively the same on the two occasions, and assuming that we hold constant all the other relevant factors pertaining to the subject's current bodily and psychological condition, there seems to be nothing in the causal influences at work which might lead to a difference in psychological outcome. But we know that, on the second occasion, the visual experience is purely hallucinatory: there is no salmon there to be seen; and, in any case, with the device activated, the subject's visual access to his external environment is automatically severed. So, if we suppose that the psychological states which are fundamentally involved on the two occasions are the same, we have to conclude that the state involved on the first occasion, in the context of seeing the salmon, is not in itself physically perceptive: it is not one whose realization, on its own, suffices for the perception of a physical item. And this forces us to say that the visual contact with the salmon, or with that momentary stage of the relevant salmon surface-portion which is Φ-terminally seen, breaks down into two components in accordance with BRT—one component consisting in Henry's being in this further, and not in itself physically perceptive, psychological state, the other comprising certain additional facts, but ones which do not involve anything further about his psychological condition at the relevant time. These additional facts will presumably, at least in part, be to do with the qualitative relationship of the state to the external environment, and with the causal role of the environment, and, in particular, of the relevant portion-stage of the salmon, in bringing about the relevant realization of this state in Henry.
Now this is just one case. But if it is sound, the reasoning involved will presumably apply with equal force to any case of physical-item perception, since, for any such case, we could envisage a similar device and tell a similar story. So we have, here, a model for a quite general argument against the
end p.24
 
SDR-account and in favour of BRT. This argument can be set out as follows:
1.  
For any physical-item perception, we can envisage a situation in which, by artificially inducing the same causally relevant neural conditions in the same subject on a subsequent occasion (these conditions covering both the process in the sensory nerves and the resulting brain response), we bring about a subjectively matching (introspectively indistinguishable) hallucination. Moreover, we can envisage this without departing from reasonable assumptions about how the world actually works.
2.  
Given the coincidence of the neural factors involved on the two occasions, and of all other simultaneous bodily and psychological factors that might be causally relevant, it is reasonable to suppose that the outcomes themselves would be, at the fundamental level of psychological description, of exactly the same psychological type.
3.  
SDR is committed to saying that the psychological state which is fundamentally involved in any instance of perception is, in itself, physically perceptive—a state whose realization, on its own, suffices for the perceiving of a physical item. For if the state were not in itself physically perceptive, the Φ-terminal perceptual relationship would have to decompose in the psychologically mediational way postulated by BRT.
4.  
But, trivially, the psychological state involved in any case of hallucination is not in itself physically perceptive, since there is no physical item which is even perceived.
5.  
So the supposition that, in the case envisaged, the two psychological outcomes would be, at the fundamental level of description, qualitatively the same commits us to rejecting SDR and offering a BRT (psychologically mediational) account of the perceptual relationship involved.
Let us refer to this as the Causal Argument from Hallucination, or, for short, the Causal Argument.
This argument may be thought to have certain echoes of a line of reasoning which I mentioned in Part One—a line of reasoning on which we
end p.25
 
passed no judgement, but which was envisaged as something which might at least tempt us to accept the narrow—classic empiricist—version of the representative theory.4 The temptation turned on two suggestions, which formed, as it were, the tentative premises of the reasoning. The first was that what enables hallucination to replicate the subjective character of physical-item perception is that, while not physically perceptive, it at least brings a real sensible object before the mind—an object which, given that it is not an ingredient of the physical world, would have to be thought of as internal to the subject's own awareness. The second suggestion was that, given that perception and hallucination have the same subjective character—that they cannot be distinguished from the standpoint of introspection—it is plausible to think of them as having the same psychological character through and through. Put together, the two suggestions invite us to accept, as the conclusion which they jointly yield, that, even in the case of perception (physical-item perception), what is immediately before the mind is an internal object of awareness, and that any contact with the external environment is channelled through the occurrence of this object in a perceptual-mediational way.
Now it is important to recognize that the Causal Argument, while indeed containing certain echoes of this line of reasoning, differs from it in two crucial respects. The first point of difference is that it is arguing for a weaker conclusion. For, instead of offering support for the narrow version of the representative theory, it only seeks to vindicate the broader (more flexible) version—the version which insists on psychological mediation, but does not impose any conditions on the nature of the mediating state involved. And, in consequence, it does not require us to accept the first of the suggestions which featured in the other line of reasoning—that hallucination involves the bringing of some real sensible object before the mind. The second point is that, instead of just relying (as in the other reasoning) on the supposed plausibility of assuming that, where mental episodes have the same subjective character, they have the same psychological character through and through, the Causal Argument focuses on a situation where there is an additional and special reason for expecting the perceptive and hallucinatory episodes to be of exactly the same psychological kind, namely that they are generated by neural conditions of exactly the same kind. This is crucial. If it were just a matter of his opponent insisting, in the abstract, that there is no room for psychological differences where there are no subjective differences, the SDR-theorist would find it easy to demur.
4 Part One, Section 2, II.
end p.26
 
And, indeed, his own account of the psychological nature of perception would give him a rationale for doing so. For if the Φ-terminal relationship is psychologically fundamental, and if the physical item perceived is external to the mind, we can hardly expect the full nature of the psychological situation, when that relationship obtains, to be introspectively transparent; and, of course, it would be blatantly question-begging just to invoke a principle of transparency as a way of excluding SDR from the start.5 But what the theorist has to accept, in a case like that of Henry, seems altogether more problematic. To maintain his account of perception, he has to insist, here, as quite generally, that the psychological states which are fundamentally involved in perception and hallucination are different (since otherwise the perceptual relationship would have to decompose in the psychologically mediational way); but he is now forced to apply this result to a case where, on the face of it, the factors which are directly causally involved in bringing about the perceptive and hallucinatory events are exactly the same. And it is this which makes his situation, prima facie, so awkward.
The difficulty, it must be stressed, is not just that what is envisaged violates the principle of same type of proximate cause, same type of immediate effect. If this were all that was involved, the SDR-theorist need not be too concerned. After all, there are other familiar cases, drawn from particle physics, where this principle, or putative principle, seems not to hold—where it seems that the same conditions can, on different occasions, give rise to different outcomes. If God can play dice with fundamental particles, why not with mental responses too? But what is particularly problematic in the present case is that the way in which this principle fails (or seems to fail) is systematic with respect to something else. For the nature of the psychological event which the central-nervous process produces does not just vary: it varies, in a regular way, with how the process itself was brought about. Thus let us continue to focus on the case of Henry. Whenever the device is switched off, the subject is able to see things in the normal way: light from the environment enters his eyes, sets up the relevant process in his central nervous system, and makes the scene before him visible. Whenever the device is activated and the nervous processes are wholly controlled by the operator, any resulting experiences are, at best, hallucinations
5 This is not to deny that such a principle (sometimes in a modified form) has been widely accepted by philosophers. Descartes, indeed, took introspective transparency to be the defining feature of the mental (cogitatio). Thus see his The Principles of Philosophy, Part 1, Principle IX, in The Philosophical Writings of Descartes, trans. J. Cottingham, R. Stoothof, and D. Murdoch, vol. I (Cambridge: Cambridge University Press, 1985).
end p.27
 
(I say 'at best' because, even to qualify as hallucinations, the experiences are required to have the subjective character of genuine perception). So, for a given relevant type of central-nervous process, the SDR-theorist has to say that this process always issues in one type of psychological event (the realization of a psychological state which is in itself physically perceptive) when it is brought about in the normal, light-induced way, and always issues in a different type of psychological event (the realization of a psychological state which is not in itself physically perceptive) when it is brought about in the artificial, device-induced way. It is this which makes the situation for the theorist so difficult. By the time the psychological event is due to be produced, there is no physical record of how the process was started. So by what mechanism does the mind adjust its response to fit the character of the remote cause? How, as it were, does the mind know whether the central-nervous process was caused in the normal or the artificial way before selecting its response? On the face of it, the right conclusion to draw is the one drawn by the Causal Argument—that the type of psychological event produced by a given type of process does not vary in accordance with how that process was itself brought about, and that, even in the case of perception, the psychological state which is fundamentally involved is one that is not in itself physically perceptive. And this would require us to say that, whenever there is perceptual contact, it breaks down into the relevant types of component in the manner stipulated by BRT.
Thus expounded, the Causal Argument looks compelling. And, as I indicated at the outset, there was a time when I thought it so myself and invoked it to justify a rejection of SDR. Subsequent reflection, however, has led me to alter my estimate of its worth. The reasons for this shift are what I shall shortly elaborate. But first, I want to take note of a quite different way in which it might be thought that the argument can be resisted.
II
In my explication of it in Part One, I drew attention to the fact that, although it takes the Φ-terminal perceptual relationship to be something psychologically fundamental, SDR does not exclude the possibility of a reductive account of mentality itself—an account which claims that psychological facts, or some relevant subset of them, are ultimately constituted by non-psychological facts.6 Now, in general, as I remarked, we should not expect the availability of this reductivist option to have much relevance to
6 See Part One, Section 3, I.
end p.28
 
the issues that concern us. After all, the dispute between SDR and BRT turns on how things are to be construed at the psychological level, and, for the most part, the considerations which bear on it are bound to be ones which arise at this level. But, with respect to the present issue, the availability of the option does, in fact, make a significant difference. For there is a way in which the SDR-theorist could pursue the option which would entirely dispose of the Causal Argument. It is not that all forms of psychological reduction would be of assistance to him here. For some—indeed, some of the most familiar—are themselves committed to BRT. This is true, for example, of the case where the reduction is effected by a behaviourist analysis of psychological states, since the psychological states involved in perception and hallucination are not distinguishable in behavioural terms. But there is one particular way of pursuing the reductive approach which is available to the SDR-theorist and which would enable him to side-step the Causal Argument altogether.
To see what this is we need to start by spelling out just why it is that the Causal Argument does not raise any problems for BRT. The reason is not that BRT does not recognize, in any sense, a psychological difference between perception and hallucination; how could anyone deny that the state of genuinely perceiving something is different from the state of merely seeming to perceive something? The reason is, rather, that BRT does not see this difference as pertaining to the states that are psychologically fundamental. Thus the BRT-theorist will say that, in the case where they subjectively match, the perceptual and hallucinatory events have the same fundamental psychological character, and that the difference in their perceptive status derives from other (non-psychological) factors—in particular, from the different ways in which the events have been brought about. So, in the case of Henry, he will say that the event of seeing the salmon and the subsequent hallucinatory episode are events of the same fundamental psychological type, but that the first qualifies as perceptive and the second as hallucinatory because of the different ways in which the optical nerve firings leading up to them have been induced—in the one case, by light from the salmon, retinally registered in the normal way, in the other, by the use of the device. There is no opportunity for a deployment of the Causal Argument because there are no differences of psychological outcome at the fundamental level of description.
Now the way in which the SDR-theorist could exploit the reductivist option is, in effect, by replicating this BRT-approach at the sub-psychological level. The account he would need to advance involves three claims. The first would be a claim of token-identity. It would assert that each mental
end p.29
 
event, or at least each mental event of the relevant (perceptual or hallucinatory) kind, is (is identical with) an event in the subject's brain. Note that, being merely an assertion of token-identity—identifying particular mental events with particular physical events—this claim does not imply that the psychological character of a relevant event is, or is an aspect of, its physical character. Nor even does it imply that, within the relevant class of events, similarities and differences in psychological and physical character are correlated. The second claim would be that the psychological character of a mental event, or a mental event of the relevant kind, is not something metaphysically fundamental, but something wholly constituted by certain physical facts about it. It is this claim which makes the relevant account reductive. Token-identity itself would allow the psychological and physical aspects of mental events to enjoy the same metaphysical status. But the additional claim of constitution represents the psychological aspects as metaphysically derivative from the physical: it represents them as obtaining in virtue of, and their obtaining as nothing over and above, the obtaining of the physical. The third claim is an elaboration of the second. It is that, within the relevant class of mental events, part of what gives a particular event its psychological character are facts about the way in which it has been brought about, and, in particular, the nature of that phase of the causal process leading up to the relevant neural process. It is this claim which allows a replication of the approach of BRT. Thus, as in the case of BRT, it allows one to say that, given an event of perceiving, part of what makes it perceptive of the relevant physical item is the fact that the causal process leads back, in an appropriate way, to that item, and correspondingly, that one way in which a mental event can be rendered hallucinatory is by there being no such physical item to which the causal process appropriately leads back. The crucial difference between the BRT and the reductive cases is that, whereas under BRT the distinction between what is derivative and what is fundamental occurs at the psychological level, on the reductive account it becomes a distinction between what obtains at the psychological level and what obtains physically. And this means that, unlike the BRT-theorist, the reductivist can insist that, at the psychological level, the states fundamentally involved in perception and hallucination are different. It is this which makes the approach available to the proponent of SDR.
It is clear why adopting this reductive account would immediately side-step the Causal Argument. On the one hand, when we focus on the situation at the metaphysically fundamental level—the level of what obtains and occurs physically—there is no difference in outcome in the perceptive
end p.30
 
and hallucinatory cases. The same type of neural process gives rise to the same type of brain event, whether the process has been brought about in the normal (perception-yielding) way, or by some form of artificial intervention, as in the case of the device. On the other hand, when we focus on how things work out psychologically, we do discern a difference in the type of outcome, and indeed one which is fundamental relative to the psychological framework of description. But the difference does not present itself as in any way problematic, since it is immediately attributable to a difference in the underlying physical factors by which the relevant psychological facts are constituted. Thus, in the case of Henry, if P is the perceptual event and H the hallucinatory, then P and H would have the same intrinsic physical character (in line with what we would expect, given the absence of any difference in the character of their proximate physical causes). But P would qualify, psychologically, as a seeing of the salmon, in virtue of the role of the salmon in bringing it about, while H would qualify as a mere hallucination, because of the absence of any physical item which plays such a role. And, in accordance with the needs of SDR, this psychological difference would count as fundamental at the psychological level of description.
There is no denying, then, that, by adopting the envisaged account, the SDR-theorist would entirely avoid the challenge of the Causal Argument: there would be no more opportunity for the deployment of such an argument in this new situation than there was in the case of BRT. What still has to be decided is whether the account is acceptable in other ways. And this promises to be a much more complex and controversial question.
One bar to any immediate endorsement of the account is that, even if we are happy with mental reductivism as such, we have not yet been given a proper idea of how the envisaged forms of constitution are supposed to work. Two areas in particular—both concerned with the perceptual case—call for a much more detailed explication of the constitutional situation if there is to be any prospect of seeing what is envisaged as acceptable.
In the first place, we would need to be told how, in the case of perception, the claim of constitution can be given a rationale without its resulting in a version of BRT. The obvious way of trying to understand how a brain event could qualify as physically perceptive in virtue of certain physical facts about it would be to see these facts as logically generating the perceptual fact in two stages. At stage one, a certain subset of the facts would suffice to endow the event with a certain representational character—with the property of representing a certain type of environmental item or situation. This stage would most likely be achieved by deploying some kind of functionalist analysis of the representational properties in question, so that
end p.31
 
an event would be thought of as having such a property in virtue of the typical functional role of events of its intrinsic type in the subject's causal system. At stage two, the event would derive its perceptive status (as the perceiving of a certain physical item) from the combination of its representational character and certain further physical facts—facts in which the properties of the relevant physical item and its causal role in bringing about the event would prominently feature. But, on the face of it, this way of trying to understand the constitutional situation would result in a version of BRT. For it seems that the assignment of a representational character to the brain event at stage one would allow us to think of the subject as being in a correspondingly representational psychological state (a state which represented a certain type of environmental item or situation, without being in itself physically perceptive); and then what is envisaged at stage two could be recast as a claim of psychological mediation, to the effect that the subject's perceiving of the relevant physical item is constituted by his being in this psychological state, together with the other physical facts. Now it may be that there is some other way of making sense of the constitutional situation—a way which sustains an SDR rather than a BRT position at the psychological level. Thus perhaps there is some way of seeing how a brain event could qualify as perceptive without giving it a representational character. Or perhaps there is some way of making sense of its representational character as something sub-psychological—so that the psychological level is only reached once the perceptive status is in place. But if there are alternatives, we would need to have them spelt out before we could begin to take the reductive proposal seriously.
The second area where there is a particular need for more elaboration relates to a point that came up earlier. When we first turned to the issue between SDR and BRT, we noted that one of the factors which makes SDR initially the more attractive option is that it is not clear how we can even make sense of what BRT envisages.7 For if the psychological states which are fundamentally involved in cases of supposed perception are not, in themselves, physically perceptive—if they are equally capable of occurring in the context of hallucination—it is hard to see how we can think of the subject's awareness as genuinely reaching to the external environment at all. Now, at the time, it seemed that SDR would not be subject to a similar worry, precisely because it postulated a perceptual relationship which was psychologically fundamental. The reaching out to the external environment was thought of as something secure in its own right, without having
7 Part Two, Section 1, II.
end p.32
 
to be sustained by more fundamental factors. But, by combining his fundamentalist thesis at the psychological level with a reductive account of the psychological facts themselves, the SDR-theorist would alter the character of the intuitive situation. For if the psychological relationship with the environment is ultimately constituted by physical facts in the way envisaged, there is the same prima facie difficulty in seeing how it can qualify as genuinely perceptual. How can we think of the subject's awareness as genuinely reaching out to the environment if, in the final analysis, the only fundamental connection between it and him is causal? What would justify the claim that the physical factors secure a genuine perceptual link, rather than simply allowing for the acquisition of environmental information? These queries may well have answers—as their counterparts may under BRT. But the answers will have to be in place before there is any chance of our finding the proposed account acceptable.
In both these ways, then—and others too—we would need to be given a much fuller account of how the reductivist approach is supposed to work before we could think of accepting it. But even if this account is available, there remains the more fundamental issue of the correctness of mental reductivism itself. Certainly, the quest for a reductive account of mentality has become very fashionable in recent years, mainly because the alternative is thought to generate insuperable problems—for example, about our epistemological access to the minds of others, and about our understanding of how mind and body causally interact. But, against this, it is also widely acknowledged that reductivism faces its own difficulties. Perhaps the chief of these is that it is hard to see how any version of it could hope to do justice to the subjective character of our mental lives—to what it feels like, from the inside, to be a conscious subject and to undergo specific forms of conscious state. And, of course, the area of perception and hallucination is a paradigm locus for the deployment of this objection, since the subjective aspects of these forms of mentality are particularly conspicuous. Now all this is a large and complex topic, and clearly one which I cannot hope to deal with, in any detail, in the context of the present investigation. If I am to be able to give proper attention to the specific topic of perception, I am forced, in this context, by and large, to put the general issues of the mind on one side. On the other hand, these issues are ones which I have discussed and tried to settle elsewhere, in a book which attempts to establish a radically dualist conception of the mind, and to discredit the various forms of reductivist and materialist alternative.8 The arguments of that book are
8 In my The Immaterial Self (London: Routledge, 1991). I should stress, as I do in that book, that a dualist conception of the mind falls short of full dualism, since it leaves open issues about the nature and status of the physical world. Thus it is compatible with mentalist and idealist accounts of the physical world.
end p.33
 
ones which I still fully endorse. So, although I cannot pursue the issue of reductivism here, I am content simply to appeal to those arguments, and draw the appropriate conclusion. As I see it, then, the reductive proposal has to be rejected, because mental reductivism of any kind can be shown to fail.
This, then, is my verdict on the reductivist approach. As it happens, it is not something on which I need to put any great weight in the context of the present discussion. If, as once, I wanted to defend the Causal Argument and use it as a weapon against SDR, then the rejection of the reductive response to it would be crucial, and this, in turn, might depend on showing that any reductive account of mentality is misconceived. But this is no longer my situation. For, as I shall shortly explain, I now think that, even when we consider the issue in purely psychological terms—leaving aside any question of the status of mentality itself—the Casual Argument fails. And although I still consider SDR to be mistaken, I shall be relying on a quite different form of argument to refute it.
The reason why I have paused to draw attention to the reductivist option is for the light it sheds on the dialectical situation. What it reveals is that the Causal Argument only has a chance of succeeding against a background in which a certain way of construing the relevant forms of mentality has already been excluded; and, in this way, it brings out a crucial limitation on what the argument on its own can hope to achieve. This limitation was not apparent at the start, when our attention was confined to how things presented themselves in the perspective of the psychological framework; and it is something which proponents of the argument are liable to miss.
III
Let us assume that the reductive response to the Causal Argument fails. The question which we must now consider is whether there is any other response which the SDR-theorist can make—one which does not require him to think of the relevant forms of mentality as reducible to something else, one which does not require him to step outside the psychological framework in which the argument itself is formulated. As we left things earlier—before our digression into the reductivist issue—the argument was looking powerful: what the theorist was obliged to envisage seemed
end p.34
 
very hard to accept. But we can hardly claim to have looked at the situation in any depth.
Two points, I think, are initially clear, and set the constraints for any form of response that might prove acceptable. First, if his position is to retain any plausibility, the SDR-theorist does indeed have to find some explanation of why the same type of neural process has different types of outcome in the two types of case. It is true that there would be nothing logically incoherent in just insisting on the difference in outcome, but leaving it unexplained, or (still more brazenly) denying that it is capable of explanation at all. But such a response would not leave his position credible. Whether or not it is ultimately conclusive, the Causal Argument reveals a major problem for his position, and, if he wants to be taken seriously, the theorist has to find some way of eliminating, or at least diminishing, it. Secondly, any satisfactory explanation of why the single type of neural process has different types of outcome in the two types of case must, in some way, represent the nature of the outcome not merely as varying with, but as depending on, the nature of the remote cause. It is true that we can envisage, in the abstract, other possibilities: for example, it could be claimed that the correlation between type of outcome and type of remote cause was the product of a pre-established harmony, Leibniz-style. But, in practice, any form of explanation other than one which postulates some kind of dependence of the nature of the outcome on the nature of the earlier causal circumstances, would not be a serious contender.
So, given these constraints, can the theorist find a satisfactory account of the situation—an effective line of defence against the argument? It seems to me that he can. In fact, the account which I have in mind is one which is already suggested by the very structure of the problem which confronts him, though its claim to merit serious consideration may not be initially apparent.
The problem, as we have said, is that the psychological outcome varies, in a regular way, according to whether the neural process which leads up to it is brought about by the normal or by the artificial method; and since the neural process does not contain any record of the way in which it was caused (otherwise it would not be exactly the same type of process in the two cases), there seems to be no room for a mechanism by which the correlation between the type of outcome and the type of remote cause could be ensured. Now, for a theorist who is determined to hold his ground, there is, in the abstract, an obvious response available. For he can say that, although the remote cause does not stamp its character on the intervening process, it causally affects the outcome directly. In other words, he can claim
end p.35
 
that, at the point when the psychological event is about to occur, the factors which directly contribute, causally, to its occurrence and character include not just the current state of the brain, but also certain aspects of the preceding causal process, including, crucially, certain aspects of the way in which the neural process leading up to the realization of the brain state has itself been brought about. This would mean that there was, after all, no violation of the principle same type of proximate cause, same type of immediate effect, since what we have hitherto been describing as the remote cause would become directly relevant to the outcome, and therefore an element of the proximate cause. More precisely, the process which leads up to the neural process would play a double causal role. It would still continue to play, in the straightforward way, the role of what brings about the neural process, and, on that score, would indeed count as the remote cause with respect to the psychological outcome. But it would also, as part of the whole physical causal process leading up to the event in the brain, combine with this event to exert a direct influence on the psychological outcome—in particular, to fix it as something perceptive or as something hallucinatory—and, on that score, would count as part of the proximate cause.
Once again, we can best illustrate what is envisaged by focusing on the pair of episodes in the case of Henry. On the first occasion, light from the salmon, and from the relevant surroundings, passes through the subject's eyes and induces a certain pattern of retinal responses; these induce a certain pattern of firings in the optic nerves, which in turn induce the realization of a certain brain state; the realization of this state is immediately followed (or perhaps accompanied) by an event of visual perception, the seeing of the salmon. On the second occasion, radio signals are transmitted by the controller and induce a certain pattern of responses in the device; these induce a pattern of firings in the optic nerves exactly like the pattern which occurred on the first occasion, and these firings in turn induce the realization of the same brain state; this is immediately followed (or accompanied) by an event of visual hallucination, which subjectively matches the earlier perception. Let us call the earlier causal process from salmon to brain event ' C 1', and the later causal process from controller to brain event ' C 2 '. What is being envisaged, then, is that the character of the psychological outcome in the two episodes is determined by the combination of the character of what occurs in the brain (which holds constant between them) and the character of the relevant causal process (which varies)—so that, on the first occasion, the realization of the relevant brain state, together with the fact that this realization has been brought about by a causal process of the C 1 -type, causes Henry to see the salmon (the psychological
end p.36
 
state involved being in itself physically perceptive), while, on the second occasion, the realization of exactly the same brain state, together with the fact that this realization has been brought about by a causal process of the C 2 -type, causes Henry to have a hallucinatory experience, as of seeing a salmon (the psychological state involved being not in itself physically perceptive). So although the relevant brain state does not preserve the information about whether its realization has been induced by the normal or by the artificial method, the very causal history leading up to its realization—including, crucially, that part of the causal history leading up to the optical-neural process which precedes it—is taken to be a directly causally relevant factor in determining what kind of psychological event ensues.
This is what the determined theorist can say. And at least there can be no denying that it offers an explanation of what needed to be explained—of why the nature of the psychological outcome varies, systematically, according to the way in which the neural process is brought about. Its prima facie drawback is that the causal mechanism it postulates, by way of explanation, seems bizarre. Everything else we know about the world suggests that causation does not work in this sort of way: it always works in a way which is not only temporally directed (from earlier to later), but temporally continuous, so that earlier events only have an influence on non-contiguous later events by affecting the chain of events that intervene. The idea that earlier events may have a direct influence on what happens after a temporal interval—an interval in which any record of the relevant features of those earlier events has been lost—seems very strange, and perhaps hardly credible. In short, it might be thought that, while the account explains the correlation, and so in that respect removes a source of puzzlement, it does so in a way which creates its own unacceptable puzzle—the puzzle of how the earlier causal history could have the sort of influence attributed to it.
But we must be careful here. It is true that, by the standards of how causation generally works, the account proposed is highly unusual. But, even before this account is imposed on it, the situation which the SDR-theorist postulates has certain distinctive features which should alert us to the possibility that our ordinary modes of causal explanation may fail to apply.
There are two relevant points here. One is simply that the area of reality to which the envisaged situation belongs is different from that which provides our ordinary paradigms of causation. These paradigms, in relation to which the postulated causal mechanism seems strange, are drawn exclusively from the physical realm: they concern the ways in which one physical event or set of conditions brings about another. But the situation for
end p.37
 
which the theorist is seeking an account concerns the relationship between the physical and the mental—between the types of physical process leading up to perception and hallucination and the nature of the psychological outcome itself. It is clearly unwarranted just to assume that forms of causality which are operative in the physical realm will be operative in the psychophysical realm as well. The psychophysical case needs to be examined on its merits.
The other point, and the more crucial one, is that, even when we put aside the problem raised by the Causal Argument, it turns out that the special form of causation envisaged, however strange-seeming in the abstract, is just what we would expect in the case of perception as the SDR-theorist construes it. This is a complex point, and we shall need to unpack it in stages.
In standard cases of causation, when some physical object is caused to come into a certain state, and when the state in question is the one that is fundamentally involved (so that the object's being in it is not constituted by its being in some further state, together with certain additional facts), this state does not consist in, or inherently include, a relationship to something earlier. In the domain of such cases—the standard cases—we would indeed be surprised, and perhaps perplexed, if we came across an instance in which (as it seemed) the object's coming into the relevant state directly causally depended not just on conditions obtaining at the time, but also on the nature of the process leading up to them. But in the case of perception, as the SDR-theorist construes it, the situation is quite different. The subject is caused to come into a psychological state which, even in its fundamental form, is inherently a relationship with an earlier item—a state which, on its own, suffices to put the subject into perceptual contact with a physical event or object-stage at an earlier time. And this does not just mean that the relevant state is one which cannot be realized without putting the subject into contact with some earlier physical item. It means that there is a particular earlier item of which the state is, in itself, perceptive; for it is precisely the Φ-terminal relationship (between the subject and the perceived physical item) which is held to be psychologically fundamental—to be what resists decomposition at the psychological level. It is hardly surprising that this aspect of the situation should make a crucial difference to our expectations about the nature of the causal process involved.
Part of the difference is contained in the merely logical point that, if a psychological state is inherently (in itself) perceptive of a particular earlier item, then nothing could directly bring about a realization of this state unless it included, or operated in a field of conditions which included, the existence of that item. For, without the existence of the item, there would
end p.38
 
be no such state to be even a candidate for realization. But there is also a more significant factor. For any subject and time, there is a multiplicity of physical items which are available for Φ-terminal perception by that subject at that time. Given that a particular item gets perceived, it is natural to suppose that there is something which causally accounts for the selection. And if we accept that the psychological state involved is inherently perceptive of the relevant item, and so accept that the existence of the item has, in any case, to be included in, or in the contextual conditions accompanying, anything which directly brings about a realization of the state, then it is natural to envisage this item as playing a direct role in that selection. It is natural to envisage it as making a direct causal contribution to the realization of the relevant state, and as doing so, in particular, in a way which ensures that the state which gets realized is perceptive of it, rather than of something else—a way which targets the subject's perceptual awareness on to itself. But not only is this what it is natural to envisage, given the theory of perception involved; it is also, crucially, just what is needed, from the standpoint of the theory, to make sense of the actual facts of human perception. After all, the form of the targeting of Φ-terminal perceptions on to physical items is, over the domain of cases, highly systematic: which item gets perceived is predictable on the basis of the environmental situation and the nature of the causal process from the environment to the subject's sensory system. For example, it is predictable that the momentary stage of a physical object which is Φ-terminally seen on any occasion is precisely that stage whose illumination is responsible for the causally relevant photic input to the eye, and thereby responsible for the causally relevant events in the subject's nervous system. Clearly this sort of regularity is not just accidental: it must have some explanation. But presumably, from the standpoint of SDR, the explanation must turn crucially on the kind of direct causal role which the relevant item plays in the targeting of the perceiving on to it. The predictability of the targeting must in some way flow from the predictability of the special influence which the item itself exerts.
But granted that the SDR-theorist has to see the perceived item as playing a direct causal role in the targeting of the perception on to it, the crucial question now is: what form would this role take? What specific contribution should we envisage the theorist as assigning to the item? Well, in part, this contribution could turn on the mere qualitative fit between the item and the content of the psychological state. For it could be that, independently of the role of the item, the other factors which directly causally bear on the psychological outcome fix the qualitative content of the state in a way which restricts its perceptive potential to objects of a certain
end p.39
 
type—a type to which the item in question conforms. But even if this qualitative restriction is part of the explanation, it cannot be the whole story, since, it will not, typically, suffice to limit the potential targets of the perceiving to just one thing. Thus while, on a given occasion, Henry Φ-terminally sees a particular portion-stage of the salmon's surface, there are obviously many other momentary items—in particular, different but qualitatively indistinguishable portion-stages drawn from the same salmon—which are qualitatively just as appropriate. And, of course, the same point would apply to almost all other cases of perception. Nor can we make up the extra factors here by merely taking account of how the various candidate-items are spatiotemporally related to the subject. For, by focusing on circumstances involving such things as mirrors or telescopes—not to mention the more complicated cases of televisions and telephones—it is easy to envisage situations where what is Φ-terminally perceived does not stand to the subject in what would normally count as the appropriate spatiotemporal relationship, and also to envisage situations where something with the appropriate character and spatiotemporal location is not what is Φ-terminally perceived. And, in any case, even where the spatiotemporal relationship involved in a perception is of the normally appropriate type (normally appropriate, that is, relative to the qualitative content of the perception and the character of the perceived item), the significance of this relationship, as a factor in the targeting, is clearly dependent on its role in facilitating the right kind of causal link. This is why, in the case of visual perception, the greater the spatial distance between the Φ-terminally perceived item and the subject, the greater the temporal distance too. Thus, when we survey the night sky, the star-stages which are currently visible to us are ones which are sufficiently earlier to have a causal impact on our current senses—even if there are other stages, of the appropriate type and in the same direction, which are temporally closer.
How then can the theorist account for the targeting? How can he see the relevant item as appropriately featuring in the conditions which causally suffice for a perception of it? The answer is now surely clear. He must invoke, and appropriately apply, the very account which we have already envisaged as his response to the Causal Argument. Thus he must say that, whenever someone Φ-terminally perceives a physical item, the nature of the psychological event which occurs directly causally depends not only on the brain state whose realization immediately precedes it, but also on the whole causal process from item to brain; and he must say that, as a crucial aspect of this direct dependence, the role of the item as the initiator of the process is (or is in the context of certain other factors) what is causally
end p.40
 
responsible for its role as perceptual object—responsible for ensuring that the resulting perception is a perception of it. In its original context, this way of envisaging the causal situation struck us as bizarre. But though it is strange by the standards of ordinary causation, it turns out to be so in a way which, quite independently of the problem raised by the Causal Argument, exactly fits, and satisfies the needs of, the distinctive nature of the psychological phenomenon whose causation is at issue. We should think of it, then, not as something which the theorist is driven to accept, as a desperate remedy against this argument, but as an integral part of the way in which he already understands the psychological nature of the perceptual relationship—as part and parcel of what is involved in his taking this relationship to be, at the point of Φ-terminality, psychologically fundamental.
In its earlier form, of course, this account was not just concerned with the case of perception, but with the case of hallucination too: it was an attempt to explain why the same type of central nervous process gives rise to different types of psychological outcome according to the way in which the process is itself brought about. But if the account is successful for the case of perception, it eo ipso has application to the case of hallucination as well. Thus if the psychological state involved in perception is always in itself perceptive of a particular physical item, and if a causally indispensable part of what directs the subject on to a particular target is the fact that a particular physical item plays a certain type of causal role in bringing about the relevant brain event, then the lack of that sort of causal factor—there not being any physical item which plays that sort of role—is bound to have a causal influence in the other direction. It is bound to ensure that, whatever the subjective character of the experience produced, no physical item is perceived—not even if there happens to be a physical item of the appropriate type in the appropriate place.
There would still be the question of what psychological nature the theorist should ascribe to such hallucinatory experiences. His theory commits him to saying that, even at the fundamental level of psychological description, they differ in character from episodes of perception. But should he take them to be merely depleted versions of what takes place in perception—a hallucinatory state coinciding with the subjective component of a perceptive state? Or should he rather think of them as involving some additional sui generis element, such as the occurrence of an internal object of awareness? The first of these approaches would, I think, be the more attractive one from the theorist's standpoint. For given that there is a wide range of quite different hallucinogenic ways of bringing about the relevant types of neural process, and that the only interesting thing which
end p.41
 
they have in common is their deviance from what is involved in the case of perception, it seems more natural to think of these ways as only having a negative influence on the psychological outcome. But this is not an issue which we need to pursue at this stage. Nor could we sensibly do so. For whatever account of hallucination the theorist adopts will have to fit in with, and be conditioned by, his account of the experiential content of perception—a topic which we shall be considering in its own right presently.
In conclusion, then, it seems to me that, notwithstanding the prima facie force of the Causal Argument, the SDR-theorist does have a reasonable line of defence, and one which he can employ without having to invoke any kind of reductive account of the relevant forms of mentality. He can explain the seemingly puzzling difference in the psychological outcomes by insisting that, contrary to initial appearances, these outcomes are directly responsive to differences in the causally relevant conditions (so that there is no departure from the expectation of same type of proximate cause, same type of immediate effect); and he can do this in a way which, even if strange by the standard of our ordinary causal paradigms, is consonant with his own distinctive view of the nature of perception. It might be retorted that, in being thus consonant, this response would just serve to bring out the method in the theorist's madness: the strangeness of his causal hypothesis would just reflect, and underline, the strangeness—unacceptable strangeness—of his conception of the perceptual relationship. But if this is so, it still needs to be demonstrated: there would have to be some independent argument to show that the relevant conception was unacceptable. And, indeed, so far from being under a prima facie cloud, this conception has, as we noted at the outset, the support of both our ordinary ways of thought and our first reflective intuitions. Clearly, the onus is on those who would reject it to point out its faults.
IV
Finally, we can now see, in retrospect, that the initial plausibility of the Causal Argument was an artefact of the way in which it was presented. The argument was formulated with a specific reference to the distinction between perception and hallucination, and, as things were presented, this focus seemed appropriate. Thus the problem for the SDR-theorist seemed to be that, on the one hand, his theory commits him to saying that the psychological states fundamentally involved in perception are in themselves perceptive, and hence different from the states involved in hallucination, while, on the other, the neural processes which lead up to the occurrence
end p.42
 
of the perceptive and hallucinatory psychological episodes could well be qualitatively identical, since the hallucinatory episodes could be brought about by inducing the normal processes in a deviant way. And it then seemed difficult for the theorist to explain the mechanics of this—the same type of neural process issuing in different types of psychological outcome, and the difference in outcome systematically correlating with the difference in the character of the remote cause. But what this presentation of the issue concealed was that the very factors which seem to create the problem are already present in the sphere of perception itself. For it is not just that SDR takes the psychological states which are fundamentally involved in perception to be in themselves perceptive (of some physical item or other). It takes them to be perceptive of particular items: it claims that, for each relevant state Σ, there is a specific physical item x, such that Σ is, in itself, perceptive (Φ-terminally perceptive) of x. And this means that, from the SDR-standpoint, the sort of situation which arises with respect to the distinction between perception and hallucination also arises, routinely, with respect to the distinction between different perceptions. For it means that, even where the causal processes from item to brain are qualitatively identical on two occasions, the psychological outcomes will be qualitatively different, simply because the two psychological states involved will be inherently perceptive of different things. If the Causal Argument had been initially viewed in this broader perspective, it would never have seemed compelling. For, as it occurs in the perceptual context, the situation does not create even a prima facie problem for the theorist. Rather, it serves to make clear the sort of causal account that his distinctive understanding of perception requires—an account which sees the whole causal process, from item to brain, as directly responsible for the nature of the perceptual outcome, and, in particular, sees the identity of the initiating item as causally fixing the identity of the perceptual target. This is an account which the theorist can accept, without embarrassment, prior to any issue over the treatment of hallucination, and can then suitably deploy, to dispose of the Causal Argument, once this issue has been raised.
3 Contact and Content
I
SDR has survived the Causal Argument. And, as we noted earlier, it will strike us, initially, as more attractive than its decompositionalist rival, having
end p.43
 
the support of both our ordinary ways of thought and our first reflective intuitions. Nonetheless, as I have already indicated, I hold the position to be mistaken, and its being so is the conclusion for which I shall now, in a series of stages, be arguing. The new line of argument will have a very different style from the one which we have been considering. Whereas the Causal Argument offered itself as a knock-down objection to SDR—focusing simply on the SDR-thesis as such—the strategy of the new argument will be to consider the range of different ways in which the thesis could be developed (the various specific SDR-accounts of perception available), and show that none of them is satisfactory. This will be a more laborious way of trying to discredit the position, but I hope, for that reason, more effective.
What I shall try to show is that, however it is developed, SDR cannot provide a satisfactory account of the experiential content of a perception, and of how this content relates to the achieving of perceptual contact. By the experiential content of a perception, I mean that content which gives the perception its subjective character—its character as introspection reveals it, or purports to reveal it. So two perceptions with the same experiential content must have the same subjective character—they must be introspectively indistinguishable. It is important, however, that we do not just equate the experiential content of a perception with its subjective character. For while the experiential content is what makes (renders) it introspectively with the subject thus and so, it does not consist in its being introspectively with him thus and so. And indeed, prior to investigation, we need to leave open the possibility of there being different types of experiential content which have the same subjective character. This is particularly important when considering a theory such as SDR. As we have already seen, the SDR-theorist is bound to think that the psychological states which are fundamentally involved in perception and hallucination are of different kinds, even when they subjectively match; and we need to leave open the possibility that this supposed difference in psychological character will involve some difference in the character of their experiential content. Indeed, it seems almost inevitable that it will.
My argument will focus on a particular form of experiential content—a form which I call 'phenomenal' content. I must start by explaining what this is.
II
Whenever someone perceives a physical item, he perceives it—or at least perceives whatever it is that he Φ-terminally perceives—under a certain
end p.44
 
sensible appearance. Thus when Pauline sees the apple, it (or the relevant Φ-terminally seen portion-stage) visually appears to her as a roughly hemispherical patch of a certain size and colouring, located at a certain distance in front of her. Likewise, when I take a bottle of milk out of the refrigerator, the bottle (or the relevant portion-stage) tactually appears to me as a cold, smooth, hard surface, of a certain curvature. Quite generally, whenever a subject Φ-terminally perceives a physical item, it sensibly appears to him in a certain way—a way which represents it as having a certain sensible character drawn from qualities associated with the relevant sense-realm, and, typically, as having a certain (though perhaps only generically specified) location relative to his own body or current position. The point of saying that the subject perceives the item under this sensible appearance is that the appearance is, on its mental side, part of the very content of the perception: the item's appearing in a certain sensible manner is one and the same as the subject's perceiving it in a certain psychological manner. Obviously, this form of perceptual content qualifies as experiential in the relevant sense.
Now it is this form of experiential content that I speak of as 'phenomenal'. In other words, the phenomenal content of the perceiving is the mental reciprocal of the item's sensible appearance: it is the item's sensible appearance taken from the other direction—what embodies this appearance in its mental aspect. It is, we might say, how, in the perceiving of that item, the subject is sensibly appeared to. Notice that, as I am understanding the notion, sensible appearance is, by definition, a form of perceptual (physical-item perceptual) appearance: it is always and necessarily the appearance of what is Φ-terminally perceived. And correspondingly, phenomenal content is, by definition, a form of perceptual (physical-item perceptual) content—always and necessarily the content of Φ-terminal perceiving. In consequence, we cannot, in my terminology, speak of there being sensible appearance or phenomenal content in the case of hallucination, even though perception and hallucination have the same subjective character. This does not rule out the possibility of saying that, at the fundamental level of psychological description, the psychological states involved in perception and hallucination are of the same kind and have experiential content of the same kind. Certainly this will be the claim of BRT. But it does mean that, whatever kind of content occurs in hallucination, we cannot, as it occurs in that context, describe it as phenomenal, or as the mental embodiment of how things sensibly appear.
Sensible appearance is exclusively a form of perceptual appearance. But not all forms of perceptual appearance count as sensible. Thus, in the case
end p.45
 
of Pauline, it may be that, as well as appearing to her as something with a certain shape, size, and colouring, the item she Φ-terminally sees appears to (is seen by) her as part of the surface of an apple; and this further aspect of how things visually appear goes beyond what is purely sensible. Likewise, the item which I Φ-terminally feel may appear to me not just as something with a certain smoothness, hardness, and curvature, but as part of the surface of a milk-bottle; and again this further aspect of how things perceptually appear transcends the purely sensible. Sensible appearance is a special kind of perceptual appearance; and correspondingly, phenomenal content is a special kind of perceptual content. Exactly what makes them special is not something which we can hope to deal with adequately at this stage: their distinctive character will only become fully clear as the outcome of our philosophical investigation. Nonetheless, there are certain distinctive features of sensible appearance which it is possible to identify in advance, without prejudging the philosophical outcome; and there are three, in particular, which I think it will be helpful to draw attention to at this point, as a way of throwing light on the investigative project.
One way in which sensible appearance differs from other forms of perceptual appearance is that the range of qualities which are available to feature in its content (which are logically capable of forming elements of how things sensibly appear) is of a narrowly circumscribed kind. Thus, for any sense-realm R, the qualities available for R-sensible appearance fall into two groups. One group, which is not, as such, distinctively linked with R, comprises qualities of a spatial or temporal kind—qualities such as shape and size, spatial patterning, subject-relative location, duration and succession. Thus the sensibly apparent shape of the relevant portion-stage Φ-terminally seen by Pauline will be drawn from this group; likewise, if someone hears a cuckoo, the sensibly apparent temporal pattern of the two sounds belongs to this group too. The other group comprises qualities which are not spatial or temporal, but which have a special relationship to R—a relationship which makes them, in some sense, distinctively R-qualities. Thus if R is the visual realm, the relevant group will comprise colours, and aspects of colour, and if R is the auditory realm, it will comprise sounds, and aspects of sound. The exact nature of the special relationship involved is not something which we can specify at this stage: it is one of the things which must await the outcome of our philosophical investigation. All I shall say at present (though here again the detailed discussion will only come later) is that where a quality stands in this special relationship to a certain sense-realm, a proper understanding of the nature of that quality involves—and distinctively involves—knowing what it is (or would be)
end p.46
 
like, subjectively, to encounter it in the experiential content of perceptions in that realm, or in the content of experiences which are subjectively like such perceptions. So it is only by knowing what it is subjectively like to encounter colours in the content of visual experiences, and sounds in the content of auditory experiences, that we can fully grasp their nature.
It should be noted that, although, in the way I have demarcated it, the first group of qualities (the spatial and temporal) does not, as a group, have any distinctive link with the relevant sense-realm, I am not excluding the possibility that some of its members have such a link, and, in particular, stand in the same special relationship to it as the qualities which occur in the second group. For example, in the case of the visual realm, I am leaving open the possibility that the group contains, amongst other things, distinctively visual qualities of shape and spatial patterning—qualities which are distinctively equipped to provide the spatial arrangement of colour, and which stand in the same special relationship to this realm as colours themselves.
Sensible appearance, then, draws its qualitative content exclusively from these two sources—from qualities of a spatial or temporal kind, and from qualities which, while not spatial or temporal, stand in the relevantly special relationship to the sense-realm involved. And this distinguishes such appearance, as a general type, from other forms of appearance, whose qualitative options are not thus restricted. It does not, however, yield a complete test of whether an instance of perceptual appearance qualifies as sensible. For although the options for non-sensible appearance are not relevantly restricted, it may happen that a non-sensible appearance draws its content exclusively from qualities of the restricted kind. For example, if someone inspects a white flower in his garden at night, when its sensible appearance is as something grey, and if he is familiar with how it sensibly looks in daylight, then the flower may still, in a certain sense, look white to him—the same sense in which, say, a ripe plum might look sweet to someone, or a pile of manure might look smelly. And this would be an instance of a non-sensible appearance, but one whose qualitative content (whiteness) is equally suitable for sensible appearance in the relevant realm. So, while the fact that an appearance draws its qualitative content partly from some source beyond the two specified groups is enough to show that it is not sensible, the fact that it does not draw on such an additional source is not enough to show that it is.
A second way in which sensible appearance differs from other forms of perceptual appearance is that it is the only form which has any essential involvement in Φ-terminal perceiving. This is so in a double sense. In the
end p.47
 
first place, it is so in a general sense, in that it is logically possible for there to be instances of Φ-terminal perceiving in which the only appearance involved is sensible, but not logically possible for there to be instances of such perceiving in which the only appearance is non-sensible. Thus we can envisage an instance of seeing in which the Φ-terminal object is merely seen under a sensible appearance—seen as an item of a certain shape, size, and colouring, in a certain subject-relative location. But we cannot envisage an instance of seeing in which the Φ-terminal object is merely seen under a non-sensible appearance—for example, as a portion or portion-stage of a certain sort of material object. For, without an underlying sensible appearance, the item would simply not be visible at all, and so would not be perceptually available as something to which some other form of appearance was attached. Secondly, the point applies to instances of Φ-terminal perceiving taken individually, in that, given any such instance, the perceptual contact with the relevant object would be unaffected if all the elements of non-sensible appearance were eliminated, but could not survive the complete elimination of sensible appearance. Thus let us assume that, in the case of Pauline, the Φ-terminal object not only (sensibly) appears to her as a roughly hemispherical patch of colour at a certain distance in front of her, but also (non-sensibly) appears to her as part of the surface of an apple. Then the non-sensible element of appearance is not needed for the perceptual contact: it could be eliminated (without replacement) without affecting her visual contact with the item itself. But the sensible appearance is crucial, since, without this appearance, or something of the same general sort in its place, the item would simply not be visible to her at all, and so would not be something which she was able to see in the relevantly apple-like way. The two applications of the underlying point are not, of course, independent. For part of what is involved in each application entails part of what is involved in the other. Thus the fact that, in each individual case, the elements of non-sensible appearance are not required for Φ-terminal contact entails that it is logically possible for Φ-terminal perceiving to occur without such elements. And conversely, the fact that it is logically impossible for Φ-terminal perceiving to occur without sensible appearance entails that, in each individual case, the sensible appearance involved plays an essential, or at least non-redundant, role.
In insisting that only sensible appearance has any essential involvement in Φ-terminal perceiving, I am allowing for the possibility of cases in which some other form of appearance is essentially involved in the securing of contact with items which are perceptually more remote—cases in which a subject's contact with one physical item is perceptually mediated by his
end p.48
 
contact with another, and where some aspect of how the more immediate object non-sensibly appears to him plays a role in the mediation. Such a possibility does not, I think, arise for the types of perceptual mediation which we have so far cited—the types exemplified by the case of Pauline. In these sorts of case—the seeing of a spatially extended object by seeing one of its parts, and the seeing of a persisting object by seeing one of its stages—the mere mereological relationship between the physical items involved seems sufficient to effect the mediation. But there are also, arguably, some quite different types of case, where it is much more plausible to think of non-sensible appearance as playing an essential role. For example, if someone detects the approach of a train by hearing the bell ring on the platform, and if we are happy to describe this as a case of actually hearing the approach of the train, it is plausible to think that his perceptual contact with this approach partly depends on the fact that he hears the bell as the signal of it—in other words, on the fact that the bell appears to him as such a signal. This is not, however, an issue which we need to settle now. The whole topic of perceptual mediation within the physical domain is something which we shall be discussing in detail later.9
There is one final distinctive feature of sensible appearance which I want to mention at this stage. It concerns the distinctive subjective character of such appearance—the distinctive introspective impression which it makes on the perceiving subject.
Let us say that an item x is directly presented to a subject S, or that S is presentationally aware of x, if and only if x is psychologically related to S in a way which satisfies three conditions. First, the relationship is such as to make x available for demonstrative identification by S. In other words, it brings x before S's mind in a way which allows him to pick it out as 'this item' (of which he is now conscious, and on to which he directs his attention), or at least in a way which would allow him to do this if he had the conceptual resources needed for such demonstrative thought (resources which he might lack if he is very young or mentally handicapped). Second, the relationship is such as to display certain aspects of x's character, or character and S-relative location, in a way which makes them immediately available for cognitive scrutiny—though, once again, S's capacity to take advantage of this availability would depend on his having the requisite conceptual resources. (The satisfying of this second condition would not, of course, be something separate from the satisfying of the first: the item would not be available for demonstrative identification in the relevant way
9 In Part Four.
end p.49
 
if there were no aspects of its character on display, and the relevant aspects would not be on display if the item were not thus available for demonstrative identification.) Third, and crucially, the relationship is wholly non-representational. It does not, in itself, involve the use of concepts, symbols, or images, as a psychological means of registering x's presence or the relevant (displayed) aspects of its character (character and location). Rather, the item and the relevant aspects are before S's mind in a mode of absolute ontological immediacy, forming, in their own ontological person, the very content of his awareness. In saying that the aspects are before S's mind in this ontologically immediate way, I mean, of course, the aspects as concretely realized—in their form as aspects of the character of x.
Now the relevant point about sensible appearance is that it carries the subjective impression of being directly presentational in this sense. It gives the subject the impression that the item he Φ-terminally perceives is before his mind in this qualitatively transparent and ontologically immediate way, with the qualitative content of the appearance exactly covering those aspects of the item's character and relative location which are seemingly displayed, and with the item and these aspects seeming to form the very content of his perceptual awareness. Thus, in as much as the portion-stage which she Φ-terminally sees sensibly appears to her as a roughly hemispherical patch of a certain size and colouring, located at a certain distance from her, Pauline is given the impression that just such an item is directly presented to her, in that locational perspective, and displaying just such qualities of shape, size, and colour. And in as much as the relevant portion-stage of the bottle sensibly appears to me as a cold, smooth, hard surface, of a certain curvature, I too am given the impression that just such an item is directly presented to me, in the relevant sort of physical contact with my body, and displaying just such qualities of temperature, texture, hardness, and shape. Moreover, this subjective aspect of sensible appearance is something which distinguishes it from other forms of appearance. Thus let us assume, as before, that what Pauline Φ-terminally sees additionally appears to her as part of the surface of an apple. Then, whatever this further appearance involves, it does not give her the impression that the item's being of this sort is something presentationally displayed—with the same kind of ontological immediacy as the display of its colour and shape. She herself, presumably, would, on introspective reflection, at once concede that, however automatic and immediate her recognition of this state of affairs, it is an aspect of how she (or how her experience) interprets the situation, rather than something which can be simply read off—as it were cognitively
end p.50
 
copied from—a pre-interpreted 'given'. Of course, on philosophical reflection, she might be led to the conclusion that sensible appearance too is, or is partly, a matter of interpretation, rather than just the transparent display of some aspect of the physical situation; and, for all we have said, such a conclusion might be right. All we are insisting on, at the moment, is that sensible appearance carries the subjective impression of being directly presentational in the relevant way, whether or not this impression is correct.
This throws light on something else. We noted earlier, as one of the factors which makes it initially attractive, that SDR is implicit in our ordinary, pre-reflective view of the nature of perception—that we ordinarily think of our perceptual awareness as reaching out to its external targets in a wholly straightforward way, which excludes any kind of decomposition at the psychological level. Given the presentational feel of sensible appearance, we can now understand why this is so. On the one hand, it is only to be expected that, prior to philosophical reflection, we will tend to think of perception in a way which accords with how it subjectively strikes us; and since perception is always under a sensible appearance, this will mean thinking of it as presentational. On the other hand, to think of perception as directly presenting its Φ-terminal objects is implicitly an endorsement of SDR. For if the Φ-terminal relationship psychologically decomposed, so that the fundamental psychological state was not, in itself, perceptive of the relevant physical item, then it would not bring this item before the mind in a way that was ontologically immediate—a way which allowed the item itself, and the relevant aspect of its character, to form the content of the subject's awareness. In short, given that perception is always under a sensible appearance, and that sensible appearance carries the subjective impression of being presentational, it is only to be expected that a presentationalist construal will come to characterize our ordinary modes of thought, and thus bring with it an implicit commitment to SDR. This is not to say that, in our ordinary thinking, we always construe perception as presentational: we are sometimes conscious of factors which show that the impression of a presentational awareness is mistaken—the obvious case being when we realize that the sensible appearance is not accurate. The point is simply that, because it coincides with how things subjectively seem, the presentationalist construal becomes established as the normal mode of construal—the construal which tends to occur in the absence of other relevant factors; and, as such, it comes to shape our ordinary basic understanding of the nature of perception prior to philosophical reflection.
end p.51
 
III
Whenever a subject Φ-terminally perceives a physical item, he perceives it under a certain sensible appearance. The phenomenal content of the perceiving is, as I have said, that aspect of its experiential content which embodies this appearance. It is the sensible appearance taken from the other direction—the manner in which the item sensibly appears, redescribed as the manner in which the subject perceives it. Any acceptable theory of perception has to be able to offer an adequate account of the nature of such content and of its relation to perceptual contact—an account of what, ultimately, phenomenal content is, and what place it occupies, or role it plays, in the obtaining of the perceptual relationship between the subject and the physical item. In one way or another, this issue will occupy our attention for most of the book. At present, my main concern is in pursuing the issue in relation to SDR, to see whether an adequate account of contact and content can be provided within its framework.
Let us focus on the issue in the context of the particular example of Pauline and the apple. Still more narrowly, let us focus on Pauline's Φ-terminal seeing, at a particular time t, of a particular momentary portion-stage, L, of the apple's surface. And, to provide for the relevant factor of phenomenal content, let us suppose that she sees L, at t, in the specific phenomenal manner M—M being what thus coincides (taken from other direction) with the manner in which L sensibly appears to her at t. So, focusing on this particular case, let us now consider what account might be given of the relevant instance of phenomenal content—of Pauline's seeing M-ly—and of how this content and the visual contact with the relevant physical item—her Φ-terminal seeing of L—fit together.
We should begin by taking note of three points that are uncontroversial. First, although Pauline at t sees L M-ly, it is logically possible for someone to see L, and to see it Φ-terminally, in a different phenomenal manner. For example, if Pauline herself had seen L from a different distance, or under a different illumination, or through shape-distorting lenses, the item would have taken on a different visual appearance and thus given rise to a different phenomenal content. From this it follows that the fact of Pauline's being in the relevant phenomenal state (of her seeing something M-ly) involves something genuinely additional to the fact of her Φ-terminally seeing L. The fact of phenomenal content is not wholly covered by the bare fact of visual contact. Let us call this Point A. Secondly, although Pauline, at t, sees L M-ly, it is logically possible for someone to see some other physical item in the same phenomenal manner. For example, if there is no
end p.52
 
change in the character of the apple or the conditions of observation, Pauline herself might well, over the next few minutes, see a whole series of successive portion-stages without any change in the character of the sensible appearance, and so without any change in the phenomenal manner of the seeing. From this it follows that the fact of Pauline's seeing L (that particular physical item) involves something genuinely additional to the fact of her being in the relevant (M-ly seeing) phenomenal state. The fact of that specifically targeted contact is not wholly covered by the bare fact of phenomenal content. Let us call this Point B. Thirdly, although the facts of contact and content are thus different—indeed, with each fact involving something genuinely additional to the other—they relate to a single concrete psychological event. It is not a matter of two things going on in Pauline's mind, simultaneously but ontologically separately—the seeing of the item and the having of a certain kind of phenomenal experience. It is a matter of one thing occurring which has both a perceptive and a phenomenal aspect—the event of seeing a particular item L in a particular phenomenal manner M. This, indeed, is simply a consequence of how the notion of phenomenal content has been defined. Let us call this Point C. These three points are, as I have said, uncontroversial, and so they are points which any philosophical account of the situation is obliged to accommodate. And, of course, although, as thus formulated, they only apply to the specific case of Pauline at t, exactly analogous points, with the same uncontroversial status, would hold for any other case of Φ-terminal perceiving.
It is not difficult to see how the three points would be accommodated by BRT. BRT claims that, whenever someone perceives a physical item, his perceptual contact with it breaks down into (is constituted by the combination of) two components, one of which consists in his being in some further, more fundamental, psychological state—a state which is not in itself perceptive of that physical item—and the other of which comprises certain additional facts, though not involving anything further (over and above his being in that state) about the subject's psychological condition at the relevant time. Applying this claim to the case of Pauline, the BRT-theorist will see both the fact of contact (Pauline's seeing of L) and the fact of content (Pauline's seeing M-ly) as decomposing into more fundamental (and the same more fundamental) factors at the psychological level. Thus he will say that there is an experiential state E, which is not in itself physically perceptive (a state which it is logically possible to be in without there being any physical item which one perceives), such that both the fact of Pauline's seeing L at t and the fact of her seeing M-ly at t are constituted by her being
end p.53
 
in E at t, together with certain additional facts of the relevantly restricted sort—facts such as the qualitative relationship of E to L and the causal role of L in bringing about the relevant realization of E. This account respects the point (Point A) that the fact of content (Pauline's seeing something M-ly) involves something genuinely additional to the fact of contact (her Φ-terminal seeing of L), since it allows for a situation in which contact with L is mediated by the realization of a different experiential state. Likewise, it respects the point (Point B) that the fact of contact involves something genuinely additional to the fact of content, since it allows for a situation in which contact with something else is mediated by the realization of the same experiential state. Finally, because it recognizes only a single fundamental psychological fact, it ensures (Point C) that there is only one concrete psychological event—an event which is, in its fundamental description, that of Pauline's being in the relevant experiential state, but which qualifies as an event of her seeing L M-ly (and so, a fortiori, as an event of her seeing L and as an event of her seeing M-ly) in virtue of the relevant additional facts.
These, of course, are just the bare bones of the BRT-account. All the details—concerning the nature of the more fundamental experiential state and the nature of the additional facts—would need to be filled in; and, for any such filling, there would still be the question of whether the account was correct—of whether it provided the right characterization of the nature of contact and content and the relationship between them. All this we shall look into in Parts Three and Four. At the moment, as I have indicated, our main concern is with the situation of SDR. Ex hypothesi, the SDR-theorist cannot take over the BRT-account, since his position denies that the Φ-terminal perceptual relationship decomposes in that kind of way. So what would be the right approach from his standpoint?
One option which is formally available—and theoretically this would be his simplest approach—would be to take the bare fact of perceptual contact, or, strictly, sense-modally specific perceptual contact, as what is psychologically fundamental, and construe the fact of phenomenal content as something constituted by the combination of contact and other factors. Let us speak of this as the perceptualist strategy. Thus, in the case of Pauline, the perceptualist would say that the bare fact of her Φ-terminally seeing L at t covers all that is relevant in her fundamental psychological condition at t, and that the fact of her seeing L M-ly at t, and a fortiori the fact of her seeing some physical item M-ly at t, are constituted by the bare fact of visual contact, together with certain additional facts—facts not involving anything further about her psychological condition at t. Obviously, this
end p.54
 
approach would be in line with SDR, since it explicitly recognizes the Φ-terminal relationship as psychologically fundamental. And it would also allow the SDR-theorist to accommodate the three uncontroversial points. In the case of Points A and C, the accommodation would be automatic: thus since it takes the content-fact to be constituted by the combination of the contact-fact and certain further facts, it automatically represents the content-fact as involving something genuinely additional to the contact-fact; and likewise, it straightforwardly ensures that there is only one psychological event—an event which is fundamentally that of the subject's perceiving the relevant item, but which qualifies as an event of perceiving in a certain phenomenal manner in virtue of the additional facts. In the case of Point B, the accommodation could be arranged by adding the appropriate claim. Thus the theorist could insist, as an essential feature of what he is envisaging, that if someone were, through the same sense-modality, Φ-terminally to perceive another relevantly similar physical item (for example, in Pauline's case, to see a different but qualitatively identical stage of the same apple), and if all the other additional factors involved were qualitatively the same, then the derivative fact of content (constituted by the fact of perceptual contact, together with the additional facts) would be qualitatively the same as well—a perceiving of a different item in exactly the same phenomenal manner. This would ensure that the fact of contact involves something genuinely additional to the fact of content to which it constitutively contributes.
The perceptualist strategy would be theoretically the simplest approach for the SDR-theorist, but it only takes a little reflection to reveal that it is misconceived. This becomes clear when we turn to the question of how it might be concretely developed. Given the bare fact of contact, what sorts of additional fact would be selected as its constitutive partners? And how would they combine with this fact of contact to yield the fact of content?
Initially, the obvious way of trying to develop the strategy would be by taking the relevant additional facts to divide into three groups, namely,
(1)  
certain facts about the intrinsic character of the perceived item;
(2)  
certain facts about the spatial or bodily-contact relationship of this item to the subject at the time of the perceiving;
and
(3)  
certain facts about the conditions of observation, over and above what is already included in (1) and (2).
end p.55
 
Thus, in the case of Pauline, the first group would cover the shape, size, and colouring of L—those intrinsic qualities of L which are capable of featuring in the content of visual sensible appearance; the second would cover the distance and direction of L (or the various points on L) from Pauline-at-t; and the third group would cover such things as the character of the illumination and the nature of the intervening medium between item and eye. The idea would be that, although there is a range of phenomenal manners compatible with the Φ-terminal perceiving of the relevant physical item in the relevant sense-realm, the fact of such perceiving determines the exact phenomenal manner when combined with these further facts.
However, there are two reasons why this proposal does not work. First, it is clear that, even if these additional facts could be thought of as partly constitutive of the subject's phenomenal condition, they do not, in conjunction with the fact of contact, suffice to determine it. Thus someone whose visual system was appropriately different from Pauline's—for example, someone who was colour-blind or astigmatic—could Φ-terminally see L with the obtaining of exactly similar additional facts, but see it in a different phenomenal manner. Secondly, the contribution which the third group of factors—the conditions of observation—make to the character of the phenomenal content is a causal, not a constitutive one: their only relevance is that they causally affect the character of the sensory input. Thus if the illumination is dim, as at dusk, this affects the phenomenal manner in which a surface is seen by reducing the quantity of light which it reflects; or if someone is looking at a scene through a pane of red glass, this affects the phenomenal manner of the seeing by filtering out some of the wavelengths of light that would otherwise reach the eye. This is a purely contingent relationship between the conditions of observation and the character of the perceptual experience, and not something which would equip the facts about these conditions to form part of what constitutes the relevant experiential fact.
The perceptualist might try to meet the first problem by supplementing the three groups of facts by a fourth group, of physiological facts—facts about the structure of the subject's sense-organs and nervous system, and about the particular neural process which led up to the perceptual event. By making these new facts sufficiently detailed, he might hope to exclude the possibility of variation in phenomenal content when all the relevant factors are duplicated. But, of course, this revision would do nothing to answer the second problem—that the conditions of observation only bear on the character of the phenomenal content causally. Indeed, it would only serve to exacerbate it, since the additional physiological factors would
end p.56
 
themselves only contribute to the character of the phenomenal content in a causal way. Admittedly, once the physiological factors had been invoked, the perceptualist could presumably arrange things in such a way that, instead of an exacerbation of the problem, there was only a transference of it to another point. For, presumably, if these new factors were sufficiently detailed to avoid the first problem, they would render any reference to the external conditions of observation redundant. But the point remains that the new facts themselves cannot be regarded as part of what constitutes the subject's phenomenal condition. At least, they cannot be so regarded outside the context of a physicalistically reductive account of the relevant psychological facts. And, even if we thought that such an account was acceptable—and I have indicated that I would reject it myself—it would not be relevant to the pursuit of the perceptualist strategy, which is seeking to characterize the situation at the psychological level.
In fact, there is no way in which the second problem for the perceptualist strategy can be avoided, and, in consequence, the strategy has to be abandoned. The only chance of success for the SDR-theorist is by adopting a quite different approach, to which I now turn.
The theorist is trying to give an account of the nature of the facts of contact and content and the relationship between them—an account which is in line with his basic position, and which allows for the three uncontroversial points that we have mentioned. On the approach we have just considered, and found wanting, he took the contact-fact (in its bare-contact form) to be psychologically fundamental, and construed the content-fact as something derivative. What he must do, instead, is to accord both these facts the same status—taking them to be different, but complementary aspects of a single fundamental fact. Thus, in the case of Pauline, he must say that what is psychologically fundamental is the fact of her Φ-terminally seeing L at t M-ly, and that the distinctive contact-fact, of her Φ-terminally seeing L (in some phenomenal manner), and the distinctive content-fact, of her Φ-terminally seeing (something) M-ly, are just different aspects of it. And he must apply this same model to all cases of Φ-terminal perception, taking the fundamental fact to be that of the subject's perceiving a particular item in a specific phenomenal manner, and treating the facts of contact and content as its different aspects. Because the unitary complex fact is what is taken to be fundamental, its relationship with the aspectual facts will be seen as one of single-fact constitution: both these latter facts will be thought of as obtaining in virtue of, and their obtaining as nothing over and above, the obtaining of the unitary fact in which they are integrated.
end p.57
 
We can now understand the significance of something which I mentioned in Part One, when first setting out the implications of the SDR-position.10 I had already noted that, in excluding the form of decomposition definitive of psychological mediation, an acceptance of SDR in effect involves taking the Φ-terminal perceptual relationship to be something psychologically fundamental—something which is not subject to any form of constitution at the psychological level. But I then added a qualification—though not one which affected the basic point. For I said that the SDR-theorist does have the option of saying that the bare fact of Φ-terminal contact is constituted by a further perceptual fact of an experientially richer or more determinate kind—a fact of which the bare contact-fact can be thought of as either an integral element or a determinable version. It is just this option which we are now envisaging the theorist as taking up, by holding the facts of contact and content to be constituted by a single fact, of which they are different aspects. The two aspectual facts can be represented as either elements or determinables of this single fact, according to how they are formulated. Thus, in the particular example, we may speak simply of Pauline's seeing L and of her seeing M-ly, which represents the aspectual facts as elements of the single fact. Alternatively, we may speak, more elaborately, of Pauline's seeing L in some phenomenal manner and of her seeing some item M-ly, which represents them as determinable versions of the single fact. Which way we care to represent the situation is purely a matter of taste. The crucial point is that it is the single fact—of perceiving a particular item in a specific phenomenal manner—which counts as psychologically fundamental.
Let us speak of this new approach as the integrational strategy, because of the way in which it preserves the integrity of the contact-content bond at the fundamental level of psychological specification. It is clear how this strategy complies with the terms of SDR—keeping the Φ-terminal relationship (specified with the requisite richness or determinacy) as something psychologically fundamental. And it is also clear how it immediately accommodates the three uncontroversial points; for each aspectual fact involves something genuinely additional to the other, while relating to a single concrete (contact-cum-content) psychological event. Moreover, with the failure of the perceptualist strategy, there is no other approach which will meet these basic requirements. Whatever our final verdict on his position, this new strategy is the only approach which offers the SDR-theorist any prospect of a satisfactory outcome.
10 In Part One, Section 3, I.
end p.58
 
4 The Presentational View
I
To have any chance of success, the SDR-theorist has to adopt the integrational strategy—taking the facts of contact and content to be aspects of a single and psychologically fundamental fact, in which they are integrated. But how exactly is this strategy to be pursued? What account should the theorist give of the two factors involved and the manner of their integration? Well, just as there was the more basic division between the perceptualist and integrational strategies, so it now turns out that, within the framework of the integrational strategy, there are two quite different approaches available to the theorist. The difference between them turns on the ontological relationship which is thought to hold between the featuring of qualities in phenomenal content, as aspects of how things sensibly appear, and their featuring in the external reality, as attributes of the item perceived. I shall begin by considering what I take to be the most obvious approach.
As we noted earlier, one of the distinctive features of sensible appearance is that it carries the subjective impression of being directly presentational, in the sense we defined. Now the theorist's most obvious approach would be to say that this subjective impression is in fact correct. That is, he could say that, whenever a subject S Φ-terminally perceives a physical item x, x is before S's mind in a way which satisfies the three specified conditions: first, it makes x available for demonstrative identification by S (as 'this thing', of which he is now conscious, and on to which he directs his attention); second, it displays certain aspects of x's character, or character and S-relative location—these aspects being what form the qualitative content of the sensible appearance; and third, it is wholly non-representational: it does not involve the use of concepts, symbols, or images, as a psychological means of registering the presence of x or the relevant (displayed) aspects; rather, x and the relevant aspects are before S's mind in a mode of absolute ontological immediacy, forming the very content of his perceptual awareness. It is this third condition which is the most crucial. Obviously, under any version of SDR, the Φ-terminal object itself must feature, in some irreducible way, in the content of the subject's perceptual awareness—otherwise the perceptual relationship would not be psychologically fundamental. But what the third condition additionally and distinctively ensures is that the content of the awareness embraces the relevant qualitative ingredients of the external situation as well. It ensures that the
end p.59
 
featuring of a quality or relation in the phenomenal content is not something ontologically separate from its external realization in the perceived item (something which merely serves to represent that realization), but is rather that realization itself made transparent to the mind—the external qualitative situation becoming experientially present, in perspective, in the content of the perceiving. So, in as much as L looks to Pauline (say) a certain shade of green, the proposal would be that this aspect of her visual experience consists in the fact that L's actual greenness, as thus externally realized, features in the content of her awareness in this ontologically immediate way. Likewise, in as much as this colour visually appears to her as distributed over a certain (roughly hemispherical) array of points, at certain distances and in certain directions from her, this aspect of her experience would be equated with the fact that the actual L-array, at such distances and in such directions, is present in the content of her awareness, with the relevant ontological immediacy, in the perspective of her spatial viewpoint. Quite generally, the proposal would be that, wherever there is Φ-terminal perception, the phenomenal content draws its qualitative ingredients directly from the concrete external situation, so that these ingredients are nothing more than elements of this situation made experientially present, in (where appropriate) the relevant perspective.
Let us call what is being proposed the presentational view. This view, as I have said, represents the theorist's most obvious approach. Part of the reason for this is that it coincides with the subjective impression which perceptual experience itself conveys—with the interpretation, as it were, which such experience puts on itself. And, in coinciding with this, it also, as we noted earlier, coincides with what we tend to take for granted in the course of ordinary (pre-philosophically-reflective) life; for our ordinary understanding of the nature of perception tends, not surprisingly, to reflect the way in which things subjectively seem when perceptual experience occurs. But there is also a further factor. For, as well as coinciding with how things subjectively seem, and with what we pre-philosophically tend to take for granted, the presentational view constitutes the simplest, most straightforward way in which the theorist could respond to the challenge of the issue before him, within the limits of what SDR and the integrational strategy allow. Thus we know that, however the situation is to be construed, phenomenal content and perceptual contact have a distinctively intimate relationship: the content is, precisely, and by definition, the phenomenal manner of the perceiving; it embodies the sensible appearance under which the contact is made. Once the BRT-approach has been excluded (so that content cannot be seen as the product of the experiential form in
end p.60
 
which the perceiving is psychologically mediated), and once the integrational strategy has been put in place, the presentational view is the simplest way of trying to accommodate that intimacy—the most straightforward way of accounting for the manner in which contact and content fit together in the relevantly integrated form. Content becomes simply the qualitative component of the presentational contact—the display of certain features of the external item or situation as part and parcel of its direct presentation.
II
The presentational view is the most obvious position for the theorist to adopt—the position which first suggests itself in the framework of the theorist's basic thesis and the integrational strategy. The trouble is that it is also—and in effect just as obviously—open to a decisive objection. The objection is that it cannot accommodate cases of non-veridical perception.
By a case of non-veridical perception, I mean one in which a physical item is genuinely perceived (in contrast with a case of hallucination), but where it is perceived in a phenomenal manner which is, in some respect, at variance with its true character—where the perceived item is not, in that respect, as it sensibly appears. That such cases occur can hardly be denied—at least on the assumption that we perceive physical items at all. A much-cited example is that of the stick in water. The stick is, in reality, straight, and remains so when it is partially immersed in water; but, thus immersed, and because of the refraction of light, it looks bent at the point where it enters the liquid, and this non-veridical look is an aspect of its sensible appearance. Another familiar example is that of the distorting effect of coloured glass. Thus if I look at newly fallen snow through red-tinted spectacles, it will look—sensibly appear—to me some shade of red, even though it is, in reality, pure white. In these examples, the distorting physical factors—the refractive effect of the immersion and the filtering effect of the glass—lie outside the subject's body. There are also examples where the relevant factors lie within—factors which affect the way in which the physical input is processed by the subject's own sensory system. For instance, there is the case of astigmatism, where the apparent shape of what is seen gets systematically affected by a defect in the lens of the eye; and there is the case of colour-blindness, where, as a result of some deficiency in the retinal receptors, certain colours which can be readily distinguished by the normal subject look the same. And, of course, for both categories of distortion—both that where the factors lie outside the subject's body and
end p.61
 
that where they lie within—there is a multitude of examples from other sense-realms, though, for the most part, I shall continue to focus on the case of sight.
Whether a perception counts as veridical or non-veridical in a given respect is not, of course, just a function of how its phenomenal content compares with the relevant qualities of the item perceived. It often depends, in addition, on the spatial relation between the subject and the item, since this can affect the requirements of perspective. Thus the correct phenomenal way of seeing a circular object from an oblique angle is not the same as the correct way of seeing it frontally; and the correct phenomenal way of seeing something in the distance is not the same as the correct way of seeing it close to. Here, though, we must be careful to distinguish between the veridical registering of a spatial quality in three-dimensional perspective and a non-veridical registering which merely captures the quality's two-dimensional projection on to the subject's viewpoint. Thus, in a situation where someone is viewing a circular object from an oblique angle, we must not confuse the case in which the orientation of the object relative to the line of vision is itself sensibly apparent, so that the different points on the object are sensibly assigned their correct values of distance, or relative distance, from the subject, and the case in which this orientation is not sensibly registered and the object appears, non-veridically, as an elliptical item directly facing one. Only the first is a case of seeing in perspective in the relevant (veridicality-preserving) sense. Nor, for similar reasons, should we suppose that the diminishing of visual acuity, which is a natural consequence of an increase in the distance of the item from the subject, is a matter of perspective. If someone does not visually register the details of a pattern because he is standing too far away from it, this cannot, except in a loose sense, be described as a case of veridically seeing what is there, but in a perspective suitable to the distance. The only cases where we can properly construe some aspect of phenomenal content as veridical-in-perspective are those where what is perceived is, in the relevant respect, quite strictly as it sensibly appears, but where the appearance records, from the standpoint of the subject (and hence in perspective), some aspect of how the item is spatially related to him.
Another thing which we need to bear in mind, when evaluating the veridicality of a perception, is that the conditions of observation may affect not just the sensible appearance of the perceived item, but its actual character in the relevant respect. Take the case of colour. When I look at snow through red-tinted glasses, this only affects the colour-appearance of the snow: its actual whiteness remains unaffected. But suppose I look at the
end p.62
 
snow at sunset, when it is illuminated by a reddish light. Again this affects the colour-appearance—perhaps in much the same way as the tinted glasses. But there is also a sense in which it affects the colour of what is illuminated too. Of course, the pigment of the snow does not change, and, in that sense, the snow remains white. But there is also a familiar and unproblematic sense in which the surface of the snow can be thought of as actually tinted by the light which falls on it, and, in that sense, its colour-appearance would be deemed veridical—an appearance in line with how things really are. A partially analogous point holds for sounds. If someone uses ear-plugs to muffle the sound of a nearby radio, we would ordinarily think of the resulting auditory perceptions as non-veridical with respect to the loudness of the physical sounds he detects: the ear-plugs serve to conceal the true intensity of these sounds, just as tinted glasses may conceal the true character of the external colours. But if the subject puts a blanket over the radio, or moves to a spot further away, we would say, I think, that he has genuinely diminished the loudness in his immediate vicinity, and so is hearing things as they now really are at his auditory viewpoint. Admittedly, in thus discussing the conditions for veridicality in respect of such qualities as colour and sound, we are implicitly assuming a common-sense account of what is involved in the physical realization of these qualities: we are assuming that the sorts of colours and sounds which feature in the content of sensible appearance are the same as the sorts which occur in the physical world. If we were to adopt a Lockean view of such 'secondary' qualities—taking them to be, in their physical realization, nothing but powers to affect human experience—then we would have to deny this assumption, and, in consequence, think of the relevant forms of sensible appearance as involving an element of non-veridicality in all circumstances.11 And, indeed, there are some familiar and seemingly powerful arguments in favour of this view. In particular, there is the argument that science is able to explain the relevant forms of appearance in terms which do not involve attributing the sensible qualities themselves to the items perceived.
In a number of respects, then, the application of the distinction between veridical and non-veridical perception is complicated, and at certain points philosophically controversial. But, for our immediate purposes, all that matters is that, however generous we may be in our recognition of cases of veridicality, there are some cases of perception which must be regarded as
11 Locke advanced this view in An Essay Concerning Human Understanding, ed. A. Campbell Fraser (New York: Dover, 1959), Bk. 2, ch. 8.
end p.63
 
non-veridical on any account. Thus, even when we make allowance for the fact that the veridical perception of shape has to vary in phenomenal manner to suit the needs of perspective, we cannot deny that a straight stick which looks bent in water does not look the way it really is, or that an astigmatic subject is liable to see physical shape in a genuinely distorted manner. And, even on a common-sense view of physical colour, we cannot deny that the effect of tinted glasses and colour-blindness is to create a real discrepancy between appearance and reality.
What makes these uncontroversial cases of non-veridicality crucial is that, on their own, they suffice to refute the presentational view. Or at least they do so if, as in the way we first formulated it, we take this view to be making a claim about all cases of Φ-terminal perception. For, in taking the qualitative ingredients of phenomenal content to be directly drawn from the external situation, the presentational view leaves no room for cases in which the character of the content—embodying how things sensibly appear—is at variance with the character of the perceived item. Phenomenal content (sensible appearance) just is, for the presentationalist, the imprint of the item's actual character, in the relevant perspective, on the percipient's mind. Full veridicality is always ensured, because it is only in so far as there is a veridical awareness of the physical world that there is phenomenal content at all. So the fact that we have to recognize cases of non-veridical perception immediately establishes that the presentational view, as a general theory of perception, is mistaken. On that point, there is no scope for dissent.
III
The fact that there are cases of non-veridical perception forces the SDR-theorist to abandon the presentational view in its general form—as a view about all cases of Φ-terminal perceiving. But it does not, immediately, force him to abandon the presentational approach altogether. After all, perception is not non-veridical all the time and in every respect. So it would still be open to him to insist that, being to some extent veridical, perception is also to some extent presentational. Let us speak of this position as the modified presentational view. This modified view, of course, leaves room for a whole range of more specific options, of varying types and strengths, depending on which forms or aspects of perception are accorded the presentational treatment. And here, there are two distinct issues whose outcomes will affect the end result. On the one hand, there is the prior issue of veridicality: in which types of case should the theorist think of perception
end p.64
 
as veridical? On the other hand, given a demarcation of the cases of veridicality, there is the subsequent issue of which of these cases should be construed in a presentational way. The strongest (i.e. presentationally strongest) version of the modified approach would be one which maximized both the extent of the veridicality and the extent of its presentational construal. It would take perception to be predominantly veridical, with non-veridicality as just a rare and minor deviation; and it would then exploit the resulting opportunities for a presentational account to the fullest possible extent.
Now it could well turn out that the correct handling of the first issue would not leave much opportunity, if any at all, for the development of a presentational account. This is partly because, even in its ordinary forms—the forms with which we are familiar from ordinary observational experience—non-veridicality is a widespread phenomenon. To appreciate its extent, we only have to remind ourselves of such common-place points as the one about acuity already mentioned (that it is often possible to get a more accurate view of the details of the physical scene by taking a literally closer look at it), and the fact that most people could, to some extent, improve their vision by wearing glasses, or a more suitable pair of glasses. But in addition, and more crucially, science seems to show that the real character of the physical world is, through and through, quite unlike how things sensibly appear in the context of ordinary observation. In part, this is just the point, already mentioned, about the status of the secondary qualities—that science seems to show that such qualities as colour, sound, flavour, and odour are nothing more, in the physical items themselves, than powers to affect human sense-experience, together with the primary structures on which these powers are grounded. But, importantly, it also extends to the primary structures themselves. For, even in respect of spatial patterning, how things sensibly appear to the ordinary visual and tactual observer is not, except in broad outline, the same as how things turn out in the light of microscopic and sub-microscopic investigation. The conclusion to be drawn seems to be that our perception of the physical world is non-veridical on a global scale, and so almost entirely beyond the reach of presentationalist treatment.
It might be suggested that we could avoid this conclusion by the device of relativization. Thus, rather than thinking of the common-sense and scientific accounts as in conflict—a conflict which, it seems, could only be rationally decided in favour of science—perhaps we should think of physical reality as having different characters relative to different perceptual and epistemological viewpoints, so that what is correctly described as a smooth
end p.65
 
and uniformly coloured surface from the viewpoint of ordinary visual observation can also be correctly described as granular and multicoloured from the viewpoint of the microscope, and ultimately described as a scattered array of colourless particles from the viewpoint of physics.12 But while this suggestion has some plausibility, it would not help to restore the opportunity for a presentational account in the relevant sense. For if veridicality really is under threat from science, and if the relativization device is the only way of preserving it in the face of this threat, then, even if we allow that, from the appropriate viewpoint, we can correctly speak of physical objects as having the character of their ordinary sensible appearance, we must also accept that their having this character, relative to this viewpoint, is, in the final analysis, wholly constituted by their possession of the properties which science ascribes to them, together with the way in which these properties dispose them to affect human sensory experience. And this would mean that, while genuinely characterizing the objects in this relativistic way, this sensible character would not be something available for presentational display at the level of what is psychologically fundamental—the only level which is relevant to the pursuit of a presentationalist approach in the framework of SDR.
It could well turn out, then, that, by the time we have finished redescribing the physical world in scientific terms, there will not be much, if anything, left in ordinary perception which is a candidate for presentational treatment. And, in fact, I am inclined to think that this is so. Moreover, we have to bear in mind that even a partial vindication of the scientific case against veridicality is, in its implications for a presentationalist approach, liable to have a knock-on effect, simply because of the ways in which qualities in one category would not be presentationally separable from those in some other. For example, once we have come to accept the Lockean view of physical colour, ruling out a presentational account of our perception of it, it surely becomes equally impossible to give a presentational account of our perception of its spatial arrangement. For even if our ways of perceiving such arrangement are (typically) veridical, it surely makes no sense to suppose that a quality of arrangement is presented on its own, without the presentation of whatever it is, in reality, that is thus arranged.
However, even if we set aside the challenge of science, and assume, quite generally, that, with well-endowed percipients in favourable conditions,
12 This seems to be P. F. Strawson's approach in 'Perception and its objects', in G. Macdonald (ed.), Perception and Identity (London: Macmillan, 1979), reprinted in J. Dancy (ed.), Perceptual Knowledge.
end p.66
 
things are, typically, as they sensibly appear, there is a quite different reason why the modified presentational view is not viable. The reason is simply that its mixing of the different approaches—its adoption of a presentational account for some cases and a non-presentational account for others—is unacceptable in principle. We can best bring this out by initially focusing on a specially designed example—one in which it is made absolutely clear that the veridical and non-veridical perceptions involved have to be treated in the same general way.
Suppose we have ten lenses, which can be arranged in a series, the first being simply a piece of plain, flat glass, of the sort one would find in an ordinary window, and the others being curved in a way which, for someone looking through them, distorts the sensible appearance of shape in a systematic fashion, the amount of the distortion being very slight initially, but steadily increasing as we move through the series. And let us assume three things about the relationship between the relevant forms of curvature and their distorting effects. First, let us assume that the different forms of curvature can be represented as different degrees of curvature along some continuous qualitative dimension, with the zero-curvature of the first lens falling at the initial point, and with the curvatures of the subsequent lenses falling at points which match their positions in the series, so that the greater the degree of the curvature involved, the greater its distorting effect. Second, let us assume that the way in which the different degrees of curvature affect the transmission of light, and so generate the phenomenal distortion, is, in its general character, constant through the series, with the effect of the zero-curvature of the first lens being just the limiting case of the effects of the curvatures of the subsequent lenses. And third, let us assume that the connections between curvature and distortion covered by these first two assumptions hold true of the dimension quite generally, or at least of the relevant portion of it, so that they do not just apply to the ten degrees of curvature that feature in our example, but also to all the other degrees that lie in between. In all these respects, of course, what is being envisaged falls securely within the framework of how the world actually works. The curvature of glass does distort the appearance of shape in the relevantly systematic way, and it would be a routine matter to construct a series of lenses which met the specified conditions.
Now, given this set-up, suppose we have a circular object O, and we have a subject who views O through each lens in turn in reverse order—each viewing being from a frontal angle. Since the final viewing is through the uncurved lens, we know that the resulting visual perception will be, in respect of O's shape, wholly veridical: the subject will see the object the way
end p.67
 
it really is—as something circular. But, in the case of the preceding nine viewings, there will be phenomenal distortion, the degree of the distortion being greatest at the start, and gradually decreasing as the viewings continue. Let us assume—and again it would be straightforward to arrange things in this way—that the distortion takes the following specific form. Throughout the first nine viewings, the object is seen, non-veridically, as elliptical, the shape of the ellipse beginning as something which is paradigmatically oval, but becoming steadily less elongated (more squat) as the series progresses, until, at the ninth (the penultimate) viewing, it is scarcely distinguishable from the circular shape which characterizes the appearance of the final perception. So while the first nine perceptions are all non-veridical, the degree of their non-veridicality steadily diminishes through the series, as the shape-appearance moves steadily closer to the appearance of circularity.
This, then, is the example on which I want to focus. And the question we need to consider is: how should we understand the nature of the perceptions which feature in it, and, in particular, how should we view the relationship between the first nine perceptions, which are all, to some degree, non-veridical, and the final perception, in which the shape of the object is seen as it really is? Well, we already know that, where a perception is non-veridical, it is not presentational: it does not, in the relevant respect, draw the qualitative ingredients of its phenomenal content directly from the character of the perceived item. And this is so irrespective of the degree of non-veridicality involved. So, in the situation envisaged, only the final—wholly veridical—perception would be a candidate for a presentational account. But we also know that, in respect of both the causation of the perceptions and the factor of veridicality, there are no sharp discontinuities. Thus all the perceptions are brought about in the same general way, the resulting shape-appearance depending, in a constant fashion, on the combination of O's actual shape and the influence of the lens on the transmission of light. And although only one perception is fully veridical, the other perceptions have a measure of veridicality which corresponds to their position in the series, so that the penultimate perception is much closer in its degree of veridicality to the final (fully veridical) perception than to those at the start. Moreover, and crucially, even where a perception is non-veridical, it is still perceptive of the object's actual shape, as physically realized—perceptive of the physical instance of circularity which forms the object's actual boundary. Thus just as the final perception involves the fully veridical seeing, or visual registering, of this physical shape, so the other perceptions likewise involve the seeing or registering of this same physical
end p.68
 
shape—this same physical instance of circularity—though in a less than fully veridical way. In this respect, the case stands in sharp contrast with what would happen if someone were viewing a circular object, but—as a variant of the case of Henry—a device which was attached to his optic nerves eliminated a certain portion of the sensory input, in a way which severed visual contact with the object's boundary and its immediate surroundings, and replaced what was eliminated with neural signals that created the false impression of an elliptical boundary. Here, the element of non-veridicality within the total visual experience would be purely hallucinatory. It would not be a case of the subject's seeing the circular boundary as elliptical, but of his not seeing the boundary at all. But, in the case of the lenses, there is no such element of hallucination. Even where a perception is non-veridical, it is, through and through, perceptive—each portion and aspect of its phenomenal content perceptually covering a real portion or aspect of the physical scene.
In view of all this, it is surely clear that all the perceptions in the series have to be thought of as of the same general kind, with the full veridicality of the final perception being just the limiting case of the increasing partial veridicality of the perceptions which precede it. It would be utterly irrational to suppose that there is a sudden and radical change in the character of the perceiving at the point where we move from the case in which, viewed through glass with only a slight degree of curvature, the object is seen (with a minute degree of inaccuracy) as almost circular, to the case in which, viewed through perfectly flat glass, it is seen (wholly accurately) as exactly circular. Such a supposition, it is true, is not logically excluded; but given the continuities in the character of the causal process and the degree of veridicality, and given that even the non-veridical perceptions are perceptive of the object's actual shape, the rational pressure to apply a uniform account throughout the series is irresistible. But this means that none of the perceptions can be taken as presentational. Even the final perception, which is fully veridical, cannot be thought of as involving the presentational display of the physical shape, since it has to be construed in the same general way as the other perceptions, whose lack of full veridicality automatically excludes a presentational account.
Now this, of course, is just one case. But it is clear that the conclusion drawn, if correct here, will extend to cases of visual perception in general. In the first place, there is obviously no theoretical significance in the fact that the object in the example was circular. If we are obliged to reject a presentational account where there is veridical viewing of a circular object through plain, undistorting glass, we are equally obliged to reject such an
end p.69
 
account for the veridical viewing of any kind of object or scene through such a medium. Secondly, there is not the slightest reason to doubt that the perceptual relationship involved in seeing something through plain, undistorting glass is of the same general character as the relationship which obtains when the only medium is air. So if we reject a presentational account for the one type of case, we have to reject it for the other as well. Finally, it is just obvious that if the presentational view fails for the straightforward case of viewing things through the medium of air—the case which would surely constitute the presentationalist's best scenario—it must be rejected for visual perception quite generally.
This result, admittedly, is still confined to the visual realm. But, although I shall not try to argue the point in detail here, similar results can be obtained for all other sense-realms too. For, in the case of each realm, it is possible to devise a similar example, in which a veridical perception is represented as the limiting case of a series of non-veridical perceptions, and in which we encounter the same irresistible pressure to apply a uniform account throughout the series. And the need to accept a non-presentational account for the veridical perception in question would then similarly extend to other veridical perceptions in that realm. In short, if our argument about the lenses has been correct, there is no way of avoiding the quite general conclusion that, in whatever form it comes, perception is never presentational.
In pressing this argument against the presentationalist, I am, admittedly, taking for granted a crucial point, namely that only the determinate aspects of the physical world are candidates for presentational display. If this point did not hold—if determinable aspects were also available for such display—then the argument would collapse. For it would then become possible to give a presentational account of veridical perception, while preserving its continuity with non-veridical. Thus, in the example of the lenses, the presentationalist could say that where the shape-appearance is (inaccurately) elliptical, there is still an aspect of O's real shape that is both veridically registered and presentationally displayed. For, taking the abstract dimension of elliptical shape, and selecting that range of determinate shapes which runs from the ellipse that characterizes the relevant appearance to the circle that characterizes O itself, he could say that what is both veridically registered and presentationally displayed is O's possession of the determinable shape-attribute which covers this range. This would allow the fully veridical final perception to provide a presentational display of O's fully determinate shape (its circularity), but in a way which gives it the same general character as the other perceptions in the series. For
end p.70
 
its display of O's circularity would be the limiting case of their display of shape-aspects of steadily increasing specificity—the increasing specificity matching the increasing degree of veridicality. The same account could then be extended to all other cases of this general kind. Thus, quite generally, the presentationalist could say that, where the perceptual registering of the presence of a determinate physical quality Q is veridical to a certain degree, it is also, to that same degree, presentational—presentationally displaying the presence of that determinable of Q which matches the measure of its phenomenal accuracy.
I have taken it for granted that only the determinate aspects of the physical world are candidates for presentational display, and so implicitly excluded this account. And the reason why I have taken this restriction for granted is that the alternative seems to me to be clearly incoherent. On the one hand, it is self-evident that a determinable attribute has no means of realization except in a determinate form. And this does not mean merely that any realization of a determinable attribute is necessarily accompanied by a realization of a determinate of it; it means that any realization of a determinable attribute is itself (qua concrete event) a realization of a determinate, and only counts as a realization of the determinable by being a realization of this determinate. Thus the only sense in which something can count as a realization of ellipticality is by being a realization of some specific elliptical shape, with precise degrees of elongation and squatness. On the other hand, the presentational display of a physical attribute is, ex hypothesi, a display of that attribute qua physically realized: the attribute features in the phenomenal content of the perception as an element of the concrete external situation, so that its featuring in that content is not something ontologically separate from its occurrence as an attribute of the physical item. In other words, what is on display is not the attribute in the abstract, but the relevant physical instance. Put together, these two points ensure that the determinable aspects of physical items are not available for presentational display in their determinable form. If they are to be available at all, it can only be as part and parcel of the availability of the determinate forms in which they concretely occur.
It turns out, then, that once the case of non-veridical perception has forced him to abandon the presentational view in its general form, the SDR-theorist is obliged to abandon this view altogether. And this is so, irrespective of how the issue about science is resolved, and, quite generally, irrespective of how broad or narrow we take the domain of veridical perception to be. The point is simply that whatever account we give of veridical perception has to apply to non-veridical perception too, or at least to
end p.71
 
the kinds of non-veridical perception on which our argument has focused. And there is no disputing that, for cases of non-veridical perception, the presentational view fails.
5 The Internalist View
I
According to the presentational view, the phenomenal content of perception directly draws its qualitative material from the physical items perceived. A sensible quality is held to feature in the content because perceptual awareness reaches out to the external environment and brings a physical realization of that quality within its scope. The featuring of the quality is not just something which serves to represent its external realization: it is that very realization made transparent to the mind—the external qualitative situation made experientially present. Now that this view has been discredited, and discredited in its entirety, the SDR-theorist is forced to think of phenomenal content in a radically different way. He is forced to say that, when a subject S Φ-terminally perceives a physical item x, and x sensibly appears to S as characterized by a certain quality Q, this featuring of Q in the phenomenal content of S's perceiving is ontologically separate from its realization in x (if it is so realized), or in anything else in the physical world—that, whatever it consists in, this phenomenal featuring of Q is not, in itself, the featuring of any physical instance of Q. In other words, although he is committed to taking the perceptual relationship with the external item to be something psychologically fundamental, and although phenomenal content is the manner in which this relationship obtains, he is forced to say that, instead of being drawn from the external environment, the qualitative ingredients of this content are wholly internal to the mind. Let us refer to this new conception of phenomenal content as the internalist view. Even if the theorist had been able to retain the presentational view in its modified form, he would, of course, have needed to adopt this internalist view with respect to those qualitative ingredients of content that resisted presentational treatment. But, having abandoned the presentational view in its entirety, he now has to embrace the internalist view in a correspondingly general form—as applying to all cases of perception and all elements of phenomenal content. So even if Pauline's seeing of L is wholly veridical—even if the way L visually appears to her, in respect of both colour and spatial qualities, exactly matches the way it is—the theorist
end p.72
 
will now have to say that the featuring of these sensible qualities in the phenomenal content of her seeing is something ontologically separate from their presence in L itself. This does not, of course, preclude him from acknowledging that the two things are causally related—from saying that it is the presence of these qualities in the relevant portion-stage which in some way makes them sensibly apparent to her. And, in acknowledging this causal relation, he could even, given a suitable conception of event-identity, think of the concrete event of phenomenal featuring as ontologically dependent on the event of physical realization; for, given the relevant conception, he could insist that the identity of the phenomenal event is partly fixed by, and logically inseparable from, the identity of its physical cause (so that there is no possible world in which that same event occurs without that same cause). But what he cannot say, as an internalist, is that the presence of a quality in this external item is what itself—in its own ontological person—features in the phenomenal content.
The SDR-theorist is forced to adopt the internalist view, and to adopt it for all cases. And there can be no denying that, in doing so, he avoids altogether the objection brought against the presentationalist. For now that the qualitative ingredients of phenomenal content are ontologically separate from their external counterparts, there is no difficulty in accommodating cases of non-veridical perception, in which sensible appearance is at variance with external reality. But this advantage, of course, does not ensure that the approach will prove to be acceptable in other ways. And, in fact, I shall argue that it has to be rejected. Specifically, I shall argue that, by embracing an internalist conception of phenomenal content, the SDR-theorist is left with no coherent account of how content and contact fit together—of how phenomenal content is able to be, what by definition it is required to be, the manner in which perceptual contact is made.
II
The first point we need to note is that, by adopting the internalist view, the SDR-theorist is taking a step in the direction of BRT. BRT claims that, wherever it obtains, the Φ-terminal perceptual relationship breaks down into two components: the realization of a psychological state which is not in itself perceptive of the relevant physical item; and certain additional facts, not involving anything further about the subject's psychological condition at the relevant time. In advancing this decompositional account, BRT is automatically committed to an internalist view of phenomenal content. But what gives it this commitment is that it is committed to an
end p.73
 
internalist account of the whole psychological state. It is committed to thinking of the qualitative ingredients of content as ontologically separate from their external counterparts because it is claiming that the perceptual relationship itself, by which the subject's awareness reaches out to the external reality, disappears at the fundamental level of psychological description. This means that, in adopting the internalist view of content, the SDR-theorist is embracing one aspect of the internalism of BRT, without embracing this internalism in its full-blooded form. He is conceding to the BRT-theorist that the qualitative ingredients of content are ontologically separate from their external counterparts, but still insisting that the perceptual relationship with the external item is a fundamental aspect of the subject's psychological condition. In this way, his position can be seen as falling between the extremes of BRT and the presentational view. BRT, as we have just said, is the position of full-blooded internalism. And, by parity of reasoning, the presentational view (the presentational version of SDR) is the position of full-blooded externalism—insisting that not only the existence of the external item, but also the relevant (sensibly apparent) aspects of its character, or character and location, fundamentally feature in the subject's psychological condition. The internalist version of SDR is a kind of compromise, or hybrid, position between these two extremes, combining the internalist element of one with the externalist element of the other.
Now it is just the fact that SDR-internalism involves this compromise which, as I see it, creates the problem. In the case of each of the extreme positions, whatever other difficulties it may face, we can see how its view about the nature of phenomenal content harmonizes, quite straightforwardly, with its view about the nature of the perceptual relationship. Thus, in the case of BRT, there may be, as we mentioned, a problem in understanding how the kind of relationship it postulates between the subject and the external environment qualifies as genuinely perceptual.13 But at least the internalist conception of the phenomenal content involved in such a relationship straightforwardly harmonizes with the nature of the relationship itself. Thus the relationship is psychologically mediated by the subject's being in some more fundamental (not in itself physically perceptive) experiential state; and, in being thus mediated, it is automatically endowed with a specific phenomenal content of an internalist kind—a content fixed by the character of the mediating state. Likewise, on the presentational version of SDR, despite the problem over non-veridicality, there is no difficulty
13 Thus see Part Two, Section 1, II.
end p.74
 
in understanding how phenomenal content and perceptual contact fit together. For the presentation of an external item would automatically involve the displaying of certain aspects of its character (or character and location) in an ontologically immediate way, and, as thus displayed, these aspects would then form, in their own person, the qualitative ingredients of the phenomenal content. But, with the internalist version of SDR, the accounts of content and contact seem, by their contrasting approaches, to pull in opposite directions. Thus if the qualitative ingredients of content are ontologically internal—ontologically separate from their external counterparts—it is hard to see how the perceptual relationship manages to be psychologically fundamental. How does the unmediated perceptual awareness manage to reach its external target if its phenomenal content does not keep (externalistic) pace with it? Conversely, if the awareness does manage to reach its external target without the content keeping pace, it is hard to see how this content can genuinely qualify as the content of the perceiving—as the manner in which the perceptual contact is made. If its qualitative ingredients stay within the boundaries of the mind—boundaries which ontologically separate the domain of the mental from the external reality—how can this content be anything other than a mere accompaniment of the perceptual relationship? In short, whatever the difficulties for the extremes, it is hard to see how there is room for the middle position. Once we adopt an internalist view of content, it is hard to see how we can understand perceptual contact in any but psychologically mediational terms; and once we adopt an SDR-conception of the perceptual relationship, it is hard to see how we can make sense of phenomenal content except as the ontologically immediate imprint of the external situation on the percipient's mind.
Now, at the moment, this only constitutes a prima facie problem for the SDR-internalist position, rather than a conclusive objection. After all, we have not yet considered, in any detail, what account of phenomenal content the SDR-internalist might want to offer. We know that, whatever that account is , it will have to represent the qualitative ingredients of content as ontologically separate from their external counterparts, and we have noted the apparent tension between this requirement and the acceptance of SDR. But it might be that the tension will disappear once the details of what is envisaged are revealed. So our next task must be to consider what options are available to the SDR-internalist, and whether any of them will serve to eliminate the difficulty which threatens to undermine his position.
One thing is already clear. Although, in adopting an internalist view of phenomenal content, the SDR-theorist is, to that extent, in agreement with
end p.75
 
BRT, he cannot just take over one of the possible BRT-accounts of content and conjoin it with his own (SDR) conception of the perceptual relationship. Under BRT, as we have said, the perceptual relationship between a subject and a physical item is psychologically mediated by the subject's being in some more fundamental, not-in-itself-physically-perceptive, psychological state, and the phenomenal content of the perceiving is then directly supplied by the character of this more fundamental state in the context of its mediational role. This means that although, to qualify for its title, phenomenal content is necessarily the content of perception (the phenomenal manner in which perceptual contact is made), the psychological phenomenon which underlies it is of a type which is not logically tied to the perceptual context: it is of a type which is equally capable of occurrence in perception and hallucination. Now it is quite clear that, if he is to have any chance of disposing of the prima facie problem, the SDR-internalist cannot simply take over an account of content of this sort and recombine it with an SDR-conception of the perceiving itself: he cannot call on the services of some category of not-in-themselves-physically-perceptive psychological states, and redeploy them as the suppliers of content for perceptual relationships which are psychologically fundamental. If a psychological state is not in itself (physically) perceptive, the only conceivable way in which it could serve as a supplier of perceptual content would be mediationally—by being a state whose realization constitutively contributed to the securing of perceptual contact in the BRT-way. What the SDR-internalist needs, then, is a form of content which is specifically and distinctively tailored to fit the claims of SDR. And this involves his thinking of it as tied to perception not merely in respect of its nominal essence (what is required for it to count as 'phenomenal'), but in its fundamental psychological nature. In other words, he has to think of it as something whose psychological character remains that of the content of perception (the manner of physical-item perceiving) at the most fundamental level of description.
The crucial question, then, is whether the theorist can find such an account of content within the constraints of his internalist view. Certainly, all the standard internalist accounts are excluded: they are ones which represent the underlying psychological state as something not in itself physically perceptive, and are, in consequence, confined to the framework of BRT. So is there any alternative? Is there any way of representing content in a form which satisfies the dual needs of the SDR-internalist position—a form which ties it to the context of perception at the fundamental level of description, while leaving its qualitative ingredients ontologically internal?
end p.76
 
If there is, it will presumably be by an account which, while leaving the qualitative ingredients internal, in some way gives them an irreducible representational concern with the character (character and location) of the relevant (Φ-terminally perceived) physical item.
As far as I can see, there is only one account which might be thought to fit the bill. This account is best seen as a crucially revised version of a theory which forms an important option under BRT. So we need to begin by outlining this BRT-theory, before introducing the changes which would bring it into line with SDR.
III
Faced with the failure of the presentational view, a number of twentieth-century philosophers have suggested that we should think of perception as making contact with the physical world, not presentationally, but by conveying information about it, and that, accordingly, we should think of phenomenal content as what embodies (or embodies the most basic part of) the information, or putative information, thus conveyed.14 This suggestion can be developed in a number of ways, according to the type of information which is thought to be involved and the supposed nature of its psychological reception. But let us refer to the generic position (however specifically developed) as the cognitive theory.
Now, notwithstanding their variation in detail, all the standard ways of developing this theory involve an acceptance of BRT. Thus, in its standard form, the generic cognitive thesis can be formulated like this:
There is a kind K of cognitive seeming, which is not in itself physically perceptive, such that, for any subject S, physical item x, and time t, if S Φ-terminally perceives x at t, then:
(1)  
it K-seems to S at t that things are currently environmentally a certain way;
(2)  
S's perceiving of x at t is psychologically mediated by his being in this cognitive state (so that the fact of perceptual contact is constituted by the fact of its K-seeming to him in the relevant way,
14 The chief exponents of this view are David Armstrong and George Pitcher. Thus see Armstrong's Perception and the Physical World (London: Routledge & Kegan Paul, 1961) and A Materialist Theory of the Mind (London: Routledge & Kegan Paul, 1968), and Pitcher's A Theory of Perception (Princeton: Princeton University Press, 1971).
end p.77
 
 
together with certain additional facts not involving anything further about his psychological condition at t);
and
(3)  
in the context of this psychological mediation, the content of the seeming logically determines the phenomenal content of the perceiving.
Thus, in the case of Pauline, the standard cognitivist would see the situation in the following terms. (i) Light from L (the relevant portion-stage of the apple's surface) enters Pauline's eyes and induces the appropriate firings in her optic nerves. (ii) These firings set up a process in the visual centres of her brain, which issues in the realization of a certain cognitive state of the relevant kind—an event of its K-seeming to her that there is a hemispherical patch of a certain size and colouring at a certain distance in front of her. (iii) Her seeing of L at the relevant time t is psychologically mediated by her being in this cognitive state: it is constituted by the combination of its K-seeming to her in the relevant way and certain further facts of the relevantly restricted kind (in particular, a suitable qualitative relationship between the content of the seeming and the actual character and subject-relative location of L, and a suitable causal role for L in bringing about this seeming). And (iv) in the context of this mediation, the phenomenal content of the seeing—L's sensible appearance as a hemispherical patch of a certain size and colouring and at a certain distance—is logically determined by the informational content of the seeming; in effect, the cognitive seeming on its own (i.e. without the perceptual contact which it mediates) covers the strictly mental aspect of how things sensibly appear—an aspect which is capable of occurring without anything being perceived—and this then qualifies as the phenomenal manner of the seeing (embodying the sensible appearance of a particular seen item) by virtue of its constitutive role, in combination with the other relevant factors, in securing contact.
It is clear how, in this standard form, the cognitive theory involves an acceptance of BRT; and, indeed, it will be as one of the major options under BRT that we shall be examining it in detail in due course. But what concerns us, at present, is that, by altering this standard version in two crucial respects, we can adapt the theory to suit the needs of SDR. And it is this revised version of the theory which might be thought to offer the SDR-internalist his one chance of success.
The first alteration is concerned with the nature of the environmental
end p.78
 
information which the theory sees as relevant. In the standard version, as outlined above, the putative information which forms the content of the relevant cognitive seeming is of a purely general character: it claims the current instantiation of a certain type of environmental situation, but does not make reference to any particular physical object or event in the subject's actual environment. So, in the case of Pauline, the putative information is merely that there is something of a certain kind and at a certain distance from her—information which makes no reference to the actual physical item seen or to any other physical particular. However, we can at least envisage a way in which the cognitivist could change this aspect of his position. For, instead of taking the relevant information to have this purely general character—claiming that things are currently environmentally thus and so—he could insist that it involves an irreducibly singular reference to the relevant physical item, around which the rest of the informational content is, as it were, ascriptively wrapped. So, in the case of Pauline, instead of saying that the way things K-seem to her is merely that there is a hemispherical patch of a certain size and colouring at a certain distance in front of her, he could say that, concerning the relevant physical item L (the particular Φ-terminally seen portion-stage of the apple's surface), the way things K-seem to her is that it—that particular item—is currently something with the relevant character and at the relevant distance.15 And he could insist that there is a similar irreducibly singular reference in the content of the K-seeming in each case of Φ-terminal perception. This, then, would constitute the first of the two relevant alterations to the standard theory. Let us speak of what results as the de re version of the cognitive theory, or, for short, the de re theory—the title 'de re' indicating the crucial role of the relevant singular reference.
Now it would be possible for a cognitivist who adopted this new version to confine the revision to that one respect. And, if he did, then the de re theory would take its place in an overall account that conformed, in an admittedly eccentric way, to BRT. Thus, despite their irreducible concern with particular physical items, the states of de re seeming would be thought of as not, in themselves, physically perceptive, and the perceptual relationship between the subject and the relevant physical item would be thought of as constituted by the cognitive relationship, together with certain further facts. But, while this would be possible, what gives the de re theory its
15 In practice, of course, the total item Φ-terminally seen by Pauline, and hence the total item to which the singular reference will need to be made, will be a larger environmental item, of which L is only one part. Confining our attention to the portion-stage of the apple's surface is purely for the sake of simplicity.
end p.79
 
interest is that it provides the opportunity for a quite different approach, involving a further and more far-reaching revision. For, instead of taking the perceptual relationship to be psychologically mediated by the cognitive relationship, the de re theorist might simply construe the perceptual relationship as this cognitive relationship, thus allowing the perceptual relationship to be something psychologically fundamental. Thus, in the case of Pauline, he might say that her Φ-terminal seeing of L under the relevant sensible appearance consists in its K-seeming to her that L (that particular item) is of the relevant character and at the relevant distance, and that because the de re seeming is something psychologically fundamental, the perceptual relationship (which he equates with it) qualifies as fundamental too. This would yield a position which accords with SDR. Moreover, as well as meeting the requirements of SDR, such a position would also meet the requirements of the internalist view. For, as on any version of the cognitive theory, the featuring of qualities in the relevant cognitive episode would be ontologically quite separate from their featuring (if they do feature) in the external situation. So the de re theory seems to offer the opportunity for an account of just the kind that the SDR-internalist needs—an account which combines, and seemingly harmonizes, a fundamentalist conception of the perceptual relationship and an internalist conception of the qualitative ingredients of phenomenal content.
IV
Have we, then, found here a satisfactory SDR-internalist account—an account which eliminates the prima facie problem and allows the theorist to reconcile the two elements of his position? Well, let me start by putting on one side two possible sources of objection.
In the first place, I shall not, in the present context, try to pursue the general issue of de re thought. There are those who would insist that the very notion of a subject's standing in an irreducibly singular cognitive relation to an external particular is simply incoherent—that in any case in which a subject stands in a cognitive relation to such an item, we can only make sense of this relational fact by taking it to decompose into the subject's being in a more fundamental cognitive state (not in itself involving such a relationship), together with certain further (non-cognitive) facts. If this claim were correct, then, of course, it would automatically rule out the viability of any de re cognitivist account of perception. But, for present purposes, I am happy to leave this challenge on one side, and proceed on the assumption that, whatever the merits or failings of the particular proposal
end p.80
 
in front of us, its commitment to the existence of de re forms of cognitive mentality is not as such objectionable.
Secondly, I shall not try to deal, at this point, with the general issue of the acceptability of the cognitive theory. There are, it must be said, a number of prima facie problems which the theory faces, irrespective of whether it is developed in the standard or de re fashion—the most obvious being the difficulty of seeing how a purely cognitive account of phenomenal content could do justice to its phenomenological character, as something quasi-presentational. This, and other general difficulties with the cognitivist approach, are ones which I shall be considering in detail in Part Three, when I focus on the cognitive theory in the framework of BRT. For the time being, I am happy, once again, to put these difficulties on one side, and leave open the possibility that some form of cognitivist account may prove acceptable.
But even when we set aside any general worries about the notion of de re thought, and any general difficulties for the cognitive theory as such, there are special reasons why we should not think of the de re theory as affording the SDR-internalist an adequate account. The main reason, as we shall see, is that, irrespective of whether it features in the context of SDR-internalism, the theory itself is simply untenable. But before I turn to this point, there is another factor which I want to bring into the picture, not only for its bearing on the particular question before us, but also because it will give us a better understanding of the full extent of the problem which the SDR-internalist faces.
V
In order for a subject to be in Φ-terminal contact with a certain physical item, the phenomenal content involved has to be, to an adequate degree, qualitatively appropriate to that item, relative to the conditions of observation. We can best bring out this point by focusing on a case where all the other conditions associated with perception are present, but the factor of appropriateness is conspicuously absent. Thus suppose that Pauline is sitting at home, with her eyes turned towards the apple on the table, with nothing obstructing her line of vision, and with all the other external factors favouring the achievement of visual contact. And suppose that light reflected from the surface of the apple, and from other things in the vicinity, enters her eyes in the normal way and sets up the appropriate kind of process in her optic nerves, which then transmit the appropriate kinds of signal to the brain. But, at this point, something peculiar happens. The
end p.81
 
brain responds to the incoming signals in a totally bizarre way, producing a visual experience which is not remotely like the sort of experience which is normal for that kind of photic input. It might be, for example, that the resulting experience is, subjectively, like that of someone viewing a herd of elephants at an African water-hole, or like that of someone seeing the Eiffel Tower in moonlight. Now it is surely clear that, given the extent of the disparity between the real character of the external situation and the content of her experience, we cannot think of Pauline as in visual contact with the apple, or anything else in her environment. It is true that the environment plays a causal role in producing the experience, and with respect to the photic input, its role is of the normal kind for those circumstances. And we can even suppose that, as in the case of normal visual perception, the brain response preserves a kind of causal isomorphism between elements of the resulting experience and elements of the input, so that, relative to a suitably fine-grained division, different elements of the experience can be causally traced back to different elements in the environment. But it would be absurd to suppose that the experience qualifies as an actual seeing of the apple, and that the only way in which its deviant content affects the situation is in making this seeing radically non-veridical. It is just obvious that, in the context of the conditions envisaged, the extent of the non-veridicality precludes visual contact altogether. So here we have a clear illustration of the point at issue, that there can only be Φ-terminal contact where there is a sufficient degree of qualitative appropriateness of phenomenal content to the item perceived, relative to the conditions of observation; and it is easy to think of a host of other examples which would illustrate this point in an analogous way. Strictly speaking, I should speak here not of the qualitative appropriateness of phenomenal content, but of the qualitative appropriateness of that experiential content which passes the subjective tests for phenomenality, i.e. that content which is either phenomenal or subjectively poses as such. For since phenomenal content is, by definition, tied to the context of (physical-item) perception, then where the degree of appropriateness fails to meet the perceptual requirements, the content involved does not qualify as genuinely phenomenal. But, having noted this point, we can conveniently continue to use the simpler, if slightly looser, mode of locution without risk of confusion.
Now we already know that, for the purposes of Φ-terminal contact, phenomenal content can have a sufficient degree of appropriateness without being fully veridical; that is why we were able to recognize cases of partially non-veridical perception, where the sensible appearance of the perceived item is in certain respects at variance with its actual character, and why, in
end p.82
 
recognizing such cases, we were obliged to reject the presentational view. But what also needs to be stressed is that the relevant kind of appropriateness is not even entirely a matter of veridicality. For it is partly a matter of conformity to whatever mode of sensible appearance is normal, or normative, for the relevant sort of item, in the relevant sort of conditions. Thus, in the case of the stick in water, there is no denying that its bent appearance is non-veridical: the stick is not in reality as it sensibly appears. But looking bent in those circumstances, where the light involved is subject to the relevant forms of refraction, is the appropriate way of looking for the purposes of perception, and it would be someone who failed to see it as bent who would need to consult an oculist. Another, and rather more complicated, example is that of colour-perception construed in a Lockean fashion—where physical colour is taken to be a mere power to produce certain kinds of colour-experience in us. For then, presumably, the appropriateness of someone's colour-experience on a given occasion will be partly a matter of conformity to the normal way in which the relevant type of pigment, or light, is disposed to look to that subject, or perhaps to some larger group of subjects to which he belongs (though it may also be thought to depend on the fact that the normal colour-system of this subject or group is sufficiently veridical with respect to degrees of qualitative similarity and difference at the physical level). However, the details of all this need not concern us. All that matters, for present purposes, is that, where there is Φ-terminal perceptual contact, there has to be an adequate degree of qualitative appropriateness, irrespective of how, precisely, such appropriateness is to be construed. Let us speak of this as the appropriateness-requirement.
There can be no denying that (in some form) the appropriateness-requirement holds. But the question we need to focus on is: why? How is the requirement to be explained? What is its rationale? After all, even in such an extreme case as that of Pauline above, there is the formal possibility of representing it as a case of perception—representing it as a case of the subject's perceiving something (in Pauline's case, the domestic scene) but under a totally deviant appearance. So exactly what is it that makes such a representation unacceptable?
From the standpoint of BRT, the situation is straightforward. For the appropriateness-requirement is an immediate consequence of the way in which perceiving is construed. Thus Φ-terminal perceptual contact is psychologically mediated by the subject's being in some more fundamental experiential state—the state which logically determines the nature of the phenomenal content—and this state's having a sufficient degree of appropriateness to the external item is a precondition of the contact being (thus
end p.83
 
mediationally) secured. In the standard version of the cognitive theory, for example, the obtaining of perceptual contact is, in part, constitutively dependent on the fact that the propositional content of the relevant seeming—its K-seeming that things are environmentally thus and so—is appropriately related to the actual environmental situation. Likewise, the requirement would be immediately explicable on the presentational view—though, of course, in a form which is distinctive to that account. For, as we have seen, by insisting that phenomenal content draws its qualitative ingredients from the external situation, the presentational view entails that wherever there is Φ-terminal perceiving, there is full veridicality. And although full veridicality is not what sufficient appropriateness actually demands, it would be what full appropriateness amounted to from the presentational view's own standpoint. But what is much more difficult to see is how the appropriateness-requirement could be explained in the framework of SDR-internalism. Ex hypothesi, the explanations offered by BRT and the presentational view are not available: the requirement cannot be represented as a consequence of the way in which perceptual contact is mediationally secured, since contact is taken to be psychologically direct; nor can it be attributed to the presentational character of perceptual awareness, since that is excluded by the internalist view of content. The difficulty is in envisaging any alternative. If contact is psychologically direct, but the qualitative ingredients of content are ontologically internal, why should there be any limit on how far the content can be out of keeping with the character of the item with which the contact is made?
This difficulty for SDR-internalism is, in a sense, just a special aspect of the general problem already identified—that the SDR-conception of the perceptual relationship and the internalist view of phenomenal content seem to be pulling in opposite directions, thus preventing any coherent account of how contact and content fit together. For obviously a proper understanding of the contact-content relationship would have to include, in particular, an understanding of why content has to reach a certain standard of appropriateness, with respect to the relevant physical item, for contact to obtain. But, while the difficulty is a special aspect of the already familiar general problem, it is not one which explicitly featured in our earlier discussion. Nor, crucially, is it one whose special significance was implicitly taken into account.
What gives this aspect its special significance is the combination of two factors. The first is simply that it presents the general problem in a particularly clear and sharp form. Even in its general form, the problem calls for a response: SDR-internalism will not be acceptable unless it can provide
end p.84
 
some account of how its externalist conception of the perceptual relationship and its internalist conception of phenomenal content can be coherently combined. But, in the specific form concerned with the appropriateness-requirement, the problem becomes especially acute. For it identifies a quite specific respect in which we know that contact and content have to harmonize, and which, on the face of it, SDR-internalism is not equipped to explain. Within the context of the general challenge to the SDR-internalist approach, what we find here is a quite clear-cut objection, which the theorist needs to address and eliminate if his position is to survive.
The second factor which gives the difficulty over the appropriateness-requirement a special significance is that it is not something which the adoption of the de re theory would, in any obvious way, help to overcome. It is easy enough to see how, assuming it to be acceptable in other respects, the theory would answer the general problem, conceived in purely general terms. For, by construing the perceptual relationship as a de re cognitive relationship, it would precisely show how this relationship can be psychologically fundamental when the qualitative ingredients of its content are ontologically internal. But what is wholly unclear is how the theory would deal with the special point about appropriateness. For how would the de re construal of the perceptual relationship help to explain why the phenomenal content—embodied by the relevant putative environmental information—stays within the limits of appropriateness which perception requires? What in the structure of the envisaged situation would exclude cases in which—taking into account all the relevant factors—the way things seem to the subject about the relevant external item is wholly out of keeping with the character of that item, relative to the conditions of observation? It seems that the only way in which the de re theorist could ensure that the putative information involved stays within the relevant limits would be by arranging for the content of this information to play a crucial role in determining that a particular physical item forms its external referent, so that the information only counts as being about a given item on condition that the properties of the item are of the relevant—appropriateness-conferring—kind. But no such arrangement can be coherently envisaged. For it is only in so far as its reference to the item is already fixed, independently of the content of the information, that it is possible for this information to be irreducibly about it; and if the information is not irreducibly about it—if it is only about it by virtue of the item's answering to some specification which the information provides—then we have shifted from the de re version of the cognitive theory to the theory in its standard form. The basic information will then simply be that there is something of
end p.85
 
a certain character which currently stands to the subject in a certain spatial relation—without singular reference to anything other than the subject and the time.16 Admittedly, there is one way in which a de re theorist could make provision for the appropriateness-requirement. For he could say that, while there is nothing in the structure of the cognitive situation which prevents the putative information from being wayward, it is only when it is sufficiently appropriate that the cognitive relationship qualifies as perceptual. But this, crucially, would be to pursue the de re approach in a BRT-fashion. It would amount to saying that perceptual contact is constituted by the combination of the cognitive relationship and those facts—about the character of the relevant item and the external situation—which, in relation to the information involved, ensure that appropriateness obtains. And so this would be of no help to someone defending a version of SDR.
It seems to me, then, that the SDR-internalist cannot deal with the problem of the appropriateness-requirement by invoking the de re theory. Nor, as far as I can see, can he deal with this problem in any way at all. As I see it, we have here a decisive objection to the SDR-internalist position, and one which is additional to the general problem that we had already encountered.
VI
If I am right, the de re theory cannot help the SDR-internalist to deal with the specific problem of the appropriateness-requirement. And if, as seems to me, there is no other source of help available, this, on its own, shows his position to be untenable. But it is not the only respect in which the outcome for his position turns out to be negative. For, on reflection, we can see that, even with respect to the general problem, the de re theory is of no assistance. And the reason for this is simply that the theory is, in itself, incoherent.
If the putative information supplied by perception is to be irreducibly de re, the de re reference must, as we noted, be fixed independently of the content of the information. But it is also necessary that the subject have, independently of his acquisition of the information, a way of grasping the reference: he must have a way of identifying the item referred to, a way of
16 From the viewpoint of the one who is receiving the information, these references would, of course, be demonstrative—the subject being identified as 'I' and the current time as 'now'.
end p.86
 
knowing which item the information concerns. After all, he is not just receiving some symbolic vehicle of information, where it would be possible to register the presence of the vehicle without registering what it signifies. The information forms the very content of his cognitive state: it is a matter of its (in some way) cognitively seeming to him that things are, with respect to the relevant item, a certain way. And this seeming would not be possible unless the identity of the item which the information is irreducibly about was part of what was cognitively registered.
The underlying point here, of course, is not specifically about perceptual information; it applies quite generally. Thus a subject cannot come, in any way, to acquire information which is irreducibly about a particular object, unless he has, independently of that information, a way of identifying the object in question. But, in the case of perceptual information—at least the perceptual information directly associated with sensible appearance—the point applies in a distinctive way. Typically, when one acquires information about some particular object, the object is something of which one already has identifying knowledge, and one is able to acquire the new information because one is able to recognize—for example, from the symbolic form in which it is being communicated—that it is that object (thus identified) which it is about. One knows already, say, who President Clinton is, and, on opening one's newspaper, learns some new fact (or alleged fact) about him—picking up the reference to the Clinton of one's prior knowledge from the occurrence of his name and title in the journalist's report. Now, in the case of perception, the subject may have, prior to the perceptual encounter, identifying knowledge of the item which he comes to perceive, or of the persisting object of which that item is a momentary stage; and, in acquiring perceptual information about it, he may then be able to exploit this knowledge and recognize what he encounters, and gains information about, as, or as a stage of, this already familiar thing. But, crucially, such prior identifying knowledge is not essential for perception; and even where it is present, the subject can only make use of it, to achieve a recognition of the identity of the perceptual object, because there is a more basic way in which, quite independently of this knowledge, the object is made available for identification by the perceptual encounter itself. This more basic form of identification is demonstrative. Thus, in order to see the apple, Pauline does not need to know of its existence beforehand. And if she does happen to have this prior knowledge (for example, she may have bought the apple the day before and placed it in its present position), her ability to apply it to the perceptual situation, and identify the item it concerns with the item which she now sees, depends on her ability to identify what she now sees
end p.87
 
in a demonstrative way, simply through its being a current object of perception.
The significance of this to our present discussion is that it puts an immediate constraint on the kind of identifiability which the de re cognitivist has to ascribe to the relevant res in the context of the relevant seeming. It means that, in taking Φ-terminal perception to involve a kind of de re cognitive seeming—a seeming which is irreducibly concerned with the particular item perceived—he has to say that the way in which the subject must be able to identify the relevant item, in order to be a subject to whom things thus seem, is the demonstrative way made available by the perceiving itself. But we can now see that the de re cognitivist is in deep trouble. For it is clear that this demonstrative mode of identification, though perceptually available, is not available independently of the item's sensible appearance. And here the point is not just that it is only available through perception, and that anything perceived is perceived under a sensible appearance. The point is that it is only in so far as an item is perceived under a sensible appearance that it is before the mind in a way which permits the relevant form of identification. Thus it is only in so far as the apple, or its relevant portion-stage, is visible to Pauline as an item of a certain shape and colouring, at a certain distance from her, that she is equipped to identify it as this thing, now perceptually before her. It is precisely through its visual appearance that it becomes relevantly identifiable.
The incoherence of the de re theory is now clear. The account presupposes that the subject has a way of identifying the item he perceives which is independent of the information which his perceiving supplies about it; while, in actual fact, the only form of identification which is needed in cases of perception—and often the only form which is available—is one which relies on the very phenomenal content which the presence of the information is (from the standpoint of the account) supposed to provide. In short, we can coherently suppose that the information fundamentally supplied by perception is de re, so long as we allow that the way in which the relevant item is perceptually identifiable is independent of the acquisition of this information. And likewise, we can coherently suppose that perception itself is to be construed in a cognitivist fashion, so long as we refrain from giving a de re account of the kind of information which features in this construal. But what we cannot coherently do is take the fundamental perceptual information to be de re, but, by construing perception itself in cognitivist terms, represent the identifiability of the relevant item as depending on the acquisition of the very information—the de re information—whose acquisition would only be possible if an independent mode
end p.88
 
of identification were available. And it is just this incoherence which we find in the de re theory.
It turns out, then, that, whatever role the SDR-internalist may have hoped to assign to it, the de re theory fails in its own terms.
VII
This completes the demise of SDR-internalism. It was clear from the start that, by combining an SDR-conception of perceptual contact with an internalist conception of phenomenal content, the theory faced a prima facie problem, since these different elements seem to pull in different directions. For a while, it seemed that the theorist might be able to deal with this problem by invoking the de re version of the cognitive theory—a version in which the information that determines phenomenal content is irreducibly concerned with a particular physical item. But, on closer examination, it turned out that the de re theory was powerless to help with the specific difficulty over the appropriateness-requirement. And it has further turned out that, whatever help it might seem to offer with respect to the general problem, the theory itself is incoherent. The upshot is that, within the framework of SDR-internalism, there is no way of providing a satisfactory account of how contact and content fit together, either in general terms, or in relation to the specific issue of appropriateness. None, at least, that I can see.
6 Conclusion
With the discrediting of SDR-internalism, our argument against SDR itself is now complete. For we have considered all the various ways in which SDR can be developed, and shown that none is satisfactory. The whole argument can be summarized thus:
1.  
SDR takes the Φ-terminal perceptual relationship to be something psychologically fundamental—in particular, something which does not decompose in the way envisaged by BRT.
2.  
Like any other theory of perception, SDR has to be able to give an adequate account of phenomenal content (the phenomenal manner of perceiving) and the way in which it relates to the obtaining of perceptual contact.
end p.89
 
3.  
One of the uncontroversial points about contact and content is that, given any case of Φ-terminal perceiving, the relevant fact of content involves something genuinely additional to the bare fact of contact, or sense-modally specific contact.
4.  
This point, together with the claim of SDR itself, limits the possibilities for an SDR-account to two broad approaches. One is the perceptualist strategy, which takes the bare fact of contact to be psychologically fundamental, and represents the fact of content as constituted by the combination of the contact-fact and certain other facts. The other is the integrational strategy, which takes the facts of contact and content to be different and complementary aspects of a single psychologically fundamental fact—the fact of the subject's perceiving a certain physical item in a certain phenomenal manner.
5.  
The perceptualist strategy can be quickly dismissed. A little reflection reveals that, in any given case, there is no set of facts which can combine with the fact of contact to yield the fact of content in the right (constitutive) kind of way. So the integrational strategy offers SDR its only chance of success.
6.  
The integrational strategy itself allows for three options, according to the ontological relationship which is held to obtain between the featuring of a quality in the phenomenal content of a perception (as a quality of how the Φ-terminally perceived item sensibly appears) and its featuring in the external environment (as a quality which the perceived item actually instantiates). First, there is the presentational view, which holds that the qualitative ingredients of content are directly drawn from the external environment, so that each phenomenal featuring of a quality is the featuring of that very instance of it which occurs in the physical item perceived (the qualitative elements of the external situation becoming present to the mind in a mode of absolute ontological immediacy). Secondly, there is the internalist view, which holds that the qualitative ingredients of content are ontologically separate from their external counterparts, so that a phenomenal featuring of a quality is not the featuring of some physical instance of it. Finally, there is the modified presentational view, which is a mixture of the other two, holding that the featuring of a quality in phenomenal content is sometimes to be construed in a presentational way, and sometimes in an internalist.
end p.90