Abstract: Let us provisionally assume a realist view of the physical world, which takes the world to be mind-independent and metaphysically fundamental. Within the framework of this assumption, there are two general theories of the nature of perception. Let us say that a subject Φ-terminally perceives a certain physical item if and only if he perceives that item, and there is no other physical item such that his perceiving of this second item mediates his perceiving of the first. Then one of the two theories endorses a realist view of the physical world, and claims that, when someone Φ-terminally perceives a physical item, his perceptual contact with that item is something psychologically fundamental. I call this 'strong direct realism' (SDR). The other theory also endorses a realist view of the world. But its claim is that, whenever someone perceives a physical item (and even when the perceiving of this item is Φ-terminal), his perceptual contact with it is psychologically mediated, i.e. it is constituted by the combination of his being in some more fundamental psychological state, which is not in itself physical-item perceptive, and certain additional facts that do not involve anything further about his current psychological condition. I call this the 'broad representative theory' (BRT). The first issue we face, then, is: which of these two theories should we accept?

Keywords: direct realism,mediation,mind-independent,perception,physical world,realism,representation

Part One The Reshaping of the Issue

1 The Traditional Issue

What is the nature of perception—the sensory perception of items in the physical world by human subjects?

By tradition, there are three general theories. First, there is direct realism. This accepts a realist view of the physical world: it takes the physical world (the world of physical space and material objects) to be something whose existence is logically independent of the human mind, and something which is, in its basic character, metaphysically fundamental. And, within this realist framework, it takes our perceptual access to the physical world to be direct. Second, there is the representative theory (or representative realism). This too accepts a realist view of the physical world. But it sees this realism—in particular, the claim of mind-independence—as putting the world beyond the reach of direct perception. Thus, in place of the claim that our perceptual access to the physical world is direct, it insists that the perceiving of a physical item is always mediated by the occurrence of something in the mind which represents its presence to us. Finally, there is idealism. This agrees with the representative theory in holding that direct perceptual awareness does not reach beyond the boundaries of the mind, but manages to combine this with the insistence that our perceptual access to the physical world is nonetheless direct. What enables it to combine these seemingly irreconcilable views is that it abandons physical realism. Thus it takes the physical world to come within the reach of direct perceptual awareness by taking it to be something which is logically created by facts about human sensory experience, or by some richer complex of facts in which such sense-experiential facts centrally feature.

Of these three traditional positions, the idealist option is likely to strike us, initially, as just absurd. This is not merely because it is an affront to 'common sense'—an outright rejection of something which we ordinarily take for granted. It is also, and more importantly, because it seems that our

end p.1

 

very concept of the physical world requires it to be something external to, and ontologically independent of, the human mind. For reasons which will emerge, I think that such a dismissal of idealism is too hasty. But, for the time being, I am going to put the idealist option on one side, and, in company with almost all other current philosophers, look at the issue of perception in the framework of physical realism.

Within this realist framework, the traditional debate focuses on the issue between direct realism, which takes our perceptual access to the physical world to be direct, and the representative theory, which takes it to be representationally mediated. However, before we can properly deal with this issue, we need to pause to reshape it. For it turns out that there are two crucially different ways in which the claim of direct access can be interpreted; and, for each interpretation, there is a corresponding version of the representative theory to stand in contrast. In other words, what poses as a unitary issue, turns out, on closer examination, to be the conflation of two distinct issues—between two distinct forms of direct realism and the correspondingly distinct forms of the representative theory. We obviously cannot hope to make any progress in our investigation until we have brought the two issues to light and decided the manner in which we are going to address them.

2 The Two Interpretations

I

We must start by getting clear about the precise point at which the different interpretations of the direct-access claim arise. And, as a first step, I need to introduce and explain two key notions: that of constitution and that of perceptual mediation. I shall take these in turn.

Let us say that a fact F is constituted by a fact F', or by a set of facts α, if and only if (1) F obtains in virtue of the obtaining of F' (the obtaining of the members of α) and (2) the obtaining of F is nothing over and above the obtaining of F' (the obtaining of the members of α). The relation of obtaining in virtue of is to be understood as necessarily asymmetric, so that we cannot envisage a case in which F obtains in virtue of the obtaining of F' and F' obtains in virtue of the obtaining of F. This means, in particular, that we cannot speak of a fact as constituted by itself (though, trivially, its obtaining is nothing over and above its obtaining); and, in effect, it means that we cannot speak of a fact as constituted by a set of facts which contain

end p.2

 

it. I am using the term 'fact' in what I take to be its standard philosophical sense, to signify an aspect of how things are. I could equally well have used the expression 'state of affairs'; my preference for 'fact' is only that it is shorter.

The definition of constitution allows it to occur in two forms, as either a relationship between two facts or a relationship between a fact and a set of facts. In effect, the distinction here is between the case where the constitutive base comprises a single fact and the case where it comprises a plurality of facts. I say 'in effect', because, technically, a set of facts could contain only one member. But to envisage a case in which a fact F is constituted by the set whose sole member is a fact F' is to envisage a case which would be more naturally described as one in which F is constituted by F'.

It is easy to find examples of both forms of constitution. A range of clear-cut cases of single-fact constitution is provided by the relationship between the instantiation of a generic (determinable) property and the instantiation of some specific (determinate) form of it. Thus if an object is scarlet, then the fact of its being (generically) red is clearly constituted, in the relevant sense, by the fact of its being (specifically) scarlet: the generic colour-fact obtains in virtue of, and its obtaining is nothing over and above, the obtaining of the more specific. A range of clear-cut cases of multi-fact constitution is provided by the relationship between the weight-relationship between two objects and their individual weights. Thus if John weighs twelve stone and Mary weighs ten stone, the fact of John's being heavier than Mary is clearly constituted, in the relevant sense, by the combination of the fact that John weighs twelve stone and the fact that Mary weighs ten stone: their weight-relationship obtains in virtue of, and its obtaining is nothing over and above, the combination of the separate weight-facts about them. Any case of multi-fact constitution can, of course, be automatically recast as a case of single-fact constitution by simply replacing the plurality of facts by their conjunction; or at least this can be done if the plurality is finite. But all this shows is that a case of single-fact constitution is only interestingly of a single-fact form if the single fact in question is not explicitly or implicitly conjunctive in that way.

Where a fact is constituted by a plurality of facts, I shall also speak of it as 'breaking down' or 'decomposing' into those facts. So, in the example above, the weight-relationship between John and Mary (John's being heavier than Mary) can be said to break down, or decompose, into the facts about their individual weights—the facts which in combination constitute that relationship.

Two other things should be noted. First, there are cases of constitutive

end p.3

 

overdetermination, where a single fact is separately constituted in two ways. For example, given any pair of facts, that p and that q, the disjunctive fact that either p or q is separately constituted by each of these facts on its own. Second, it will be convenient to construe the relation of obtaining in virtue of, as it features in the definition of constitution, in a way which excludes forms of redundancy in the constitutive base. Thus if F obtains in virtue of the obtaining of F', then we are not to speak of F as obtaining in virtue of the obtaining of the conjunction of F' and some further fact, or in virtue of the obtaining of some plurality of facts which includes F'. And if F obtains in virtue of the obtaining of the members of α, we are not to speak of it as obtaining in virtue of the obtaining of the members of some larger set which includes α. This means, in particular, that, in cases of constitutive overdetermination, we are not to speak of the fact which is doubly constituted as also constituted by the combination of the facts, or sets of facts, which are separately constitutively involved.

With the notion of constitution in place, I can now explain the notion of perceptual mediation. It often happens that the perceiving of one thing is wholly channelled through the perceiving of another. This occurs when, for some subject S, time t, and items x and y,

(1)	S perceives x at t;
(2)	S perceives y at t;
(3)	S's perceiving of x at t breaks down into (is constituted by the combination of) his perceiving of y at t and certain additional facts; and
(4)	these additional facts do not involve anything further about S's perceptual condition at t (anything over and above what is already covered by S's perceiving of y). In other words, in combining with the fact of S's perceiving of y, they do not add further perceptual facts, about S at t, to the constitutive base.

It is this phenomenon that I call 'perceptual mediation'; and, in cases where it obtains, I speak of the perceiving of the relevant y as perceptually mediating the perceiving of the relevant x, and the perceiving of x as perceptually mediated by the perceiving of y.

Perceptual mediation can assume a number of forms (or putative

end p.4

 

forms) according to the nature of the additional facts involved—the facts which combine with the perceiving of the one item to constitute the perceiving of the other. We shall be looking at these different forms in due course. At present, it is best to confine our attention to a very simple type of case. Thus suppose Pauline is looking at an apple on the table. She sees the apple (A), and, in that sense, sees the whole apple. But she does not see the whole of (every part of) A: the only part which is strictly visible to her is a certain portion (P) of A's surface. In consequence, her seeing of A breaks down into (is constituted by the combination of) her seeing of P and the fact that P is a portion of A's surface. And this means that her seeing of P perceptually mediates her seeing of A in the sense defined. We can take the example one stage further. The relevant portion of A's surface, like A itself, is a temporal continuant—something which persists through time. But, at each moment when Pauline sees this portion, she sees it, not as it is at the various stages of its history, but only as it is at that moment—or strictly, given the time it takes for light to travel from the surface of the apple and get visually registered, she sees it as it was at a fractionally earlier moment.¹ This means that, just as Pauline's seeing of A is perceptually mediated by her seeing of P, so, at any particular moment, her seeing of P is perceptually mediated by her seeing of a certain momentary stage (time-slice) of it. For it means that, at each such moment, her seeing of P breaks down into (is constituted by the combination of) her seeing of the relevant momentary item and the fact that this item is a momentary stage of P. Obviously, we could apply this same two-stage analysis to any instance of seeing a material object—to yield any number of further cases in which the seeing of the whole object is perceptually mediated by the seeing of a part, and the seeing (at a particular moment) of this part is perceptually mediated by the seeing of a time-slice.

In considering cases of perceptual mediation, it must always be borne in mind that the relationship holds between facts of perceiving, not acts (concrete events) of perceiving. Thus when we say that a subject's perceiving of one thing is perceptually mediated by his perceiving of another, we are saying that the fact of his perceiving the first thing is constituted by the fact of his perceiving the second, together with certain additional facts of the relevant sort. What makes this point particularly crucial is that, in cases of perceptual mediation, the relevant acts (concrete events) of perceiving will often be the same. Indeed, they will always be the same in cases where the

¹ Even this is an over-simplification, since the distance from surface to visual system is not constant over surface-points.

end p.5

 

additional facts do not involve anything further about the subject's psychological condition at the relevant time. Thus, in the case of Pauline, while the fact of her seeing A is perceptually mediated by the fact of her seeing P, and the fact of her seeing P is in turn perceptually mediated by the fact of her seeing the relevant P-stage, there is only one act (one concrete event) of seeing involved. Thus the act of seeing the momentary item qualifies as an act of seeing P, in virtue of the item's being a stage of P; and the act of seeing P (which is itself identical with the act of seeing the P-stage) qualifies as an act of seeing A, in virtue of P's being a portion of A's surface.

II

The direct realist claims that our perceptual access to the physical world is direct. Now, in making this claim, he is not wanting to deny that there are cases of perceptual mediation of the sort that we have just been considering. He is not wanting to deny that, in a case like Pauline's, the seeing of the whole object is perceptually mediated by the seeing of the part, and the seeing of the persisting part is perceptually mediated by the seeing of the relevant stage. Nor is he wanting to deny that there is a variety of other types of case, which we have not yet discussed, in which the perceiving of one physical item is perceptually mediated by the perceiving of another. What he is wanting to claim is that our perceptual contact with things in the physical world becomes direct at the point where there is no further perceptual mediation within the physical domain. Thus let us say that a subject S Φ-terminally perceives an item x if and only if x is a physical item and S perceives x and there is no other physical item y such that S's perceiving of x is perceptually mediated by his perceiving of y. Then the direct realist's claim is exclusively concerned with the nature of physical-item perceiving at the point of Φ-terminality. His claim is that Φ-terminal perceiving—perceiving not subject to any further perceptual mediation within the physical domain—is direct. In making this claim, he is assuming that, in any case of physical-item perception, there is a point of Φ-terminality—that the series of physical-domain mediational links is not, in the direction of the perceiver, infinitely regressive. But this assumption is surely safe enough. In the case of Pauline, indeed, which represents a whole class of analogous cases, we have already, it seems, identified the Φ-terminal point. For there is surely no physical item which she sees more immediately than the relevant time-slice of the portion of the apple's surface.

It is at this point that the different interpretations of the direct realist position arise. For the claim that Φ-terminal perceiving is direct can be

end p.6

 

understood in two ways. On each construal, the claim is concerned with excluding, at the point of Φ-terminality, a certain form of mediation. But the two forms of mediation involved—the two senses in which the perceptual relationship is taken to be unmediated—are crucially different.

In the context of our recent discussion, the most obvious way of understanding the claim of directness would be as excluding any further form of perceptual mediation—as claiming that where someone Φ-terminally perceives a physical item x, there is no further item y such that his perceiving of y perceptually mediates his perceiving of x. In other words, if we speak of perceiving which is not subject to perceptual mediation as perceptually direct, then the most obvious way of understanding the claim would be as a claim of perceptual directness in this sense. Such a claim would not be trivial. For although, by definition, Φ-terminal perceiving cannot be perceptually mediated by the perceiving of something physical, it does not follow that it cannot be perceptually mediated by the perceiving of something non-physical—something which is not an ingredient of the world of physical space and material objects.

Thus interpreted, the claim of directness is not trivial. But it will only be of philosophical interest if there is some identifiable position which we might think of adopting and which the claim excludes—some definite way in which we might come to suppose that perceptual contact with the physical world involves this further stage of mediation. And, in fact, there is a major tradition in the philosophy of perception which espouses just such a view. For it holds that whenever a subject perceives a physical item, his perceiving of it is perceptually mediated by an awareness of something which only exists in his own mind. At first sight, such a view is likely to strike us as very strange: it does not ordinarily occur to us that our access to the physical world might be mediated in that sort of way. Nonetheless, there are a number of considerations which could lead us to take the view seriously, and although this is not the moment to discuss them in any detail, it will be helpful to mention one in particular, as a way of shedding a little more light on the content of the view itself. This consideration is concerned with the phenomenon of hallucination.

In cases of hallucination, or at least the kind of hallucination that presently concerns us, the subject has an experience which is subjectively just like that of perceiving a physical item, though without there being any physical item which is perceived—an experience which is not physically perceptive, but which is introspectively indistinguishable from one which is. Now it is at least tempting to say that a crucial part of what enables the hallucinatory experience to replicate the subjective character of an

end p.7

 

ordinary perception is that, although not physically perceptive, it does genuinely bring some sensible object before the mind. Thus, in the case of visual hallucination, it is tempting to say that a crucial part of what enables it to replicate the subjective character of ordinary (physical-item) seeing is that, although nothing physical is seen, there is a real array of colours visually before the mind—an array which poses as, and invites the subject to take it to be, an ingredient of the physical world. Likewise, in the case of auditory hallucination, it is tempting to say that a crucial part of what enables it to replicate the subjective character of ordinary (physical-item) hearing is that, although nothing physical is heard, there is a real sound or complex of sounds auditorily before the mind—an item which, again, poses as, and invites the subject to take it to be, an ingredient of the physical world. But if there are indeed such qualitative items before the mind—items which, ex hypothesi, are not themselves ingredients of the physical world—they will presumably have to be entities which are internal to the subject's awareness—entities which have, and can have, no existence outside the context of the awareness directed on to them. And once we have come to accept the presence of these internal objects of awareness in the case of hallucination, it then becomes tempting to recognize their presence in the case of perception—physical-item perception—too. For if perception and hallucination have the same subjective character, the simplest and most obvious way of accounting for this would be to suppose that they also have, at the fundamental level of description, the same psychological character through and through, and that what distinguishes the two types of case is simply that, where the experience is physically perceptive, it stands in some appropriate qualitative and causal relationship to the external environment. In this way, we could be led to the conclusion that, even in the case of physical-item perception, what is immediately before the mind is an object which is internal to the subject's awareness, and that contact with the external environment is achieved by perceptual mediation—as something constituted by the perceiving of this internal object, together with additional facts of the relevant kind. The basic idea would be that, by standing in the appropriate qualitative and causal relations to a certain physical item, the internal object would serve to represent it—in something like the way in which a photograph can serve to represent an earlier photographed scene, or the playing of a tape can serve to represent an earlier recorded conversation—and that this representation would then suffice to put the subject and the item into perceptual contact.

This position constitutes the representative theory of perception in its classic empiricist form, and, from Locke onwards, it has been widely

end p.8

 

endorsed by philosophers in the empiricist tradition. Some of those who have endorsed it have preferred to restrict the terms 'perception' and 'perceptual awareness' to the relationship between the subject and the physical item, and to use such terms as 'sensing' and 'sensory awareness' for the subject's relationship to the items in his mind. And so, while accepting that the mental items are what the subject is immediately aware of, they prefer to avoid speaking of them as what he immediately perceives. But this is just a variation in terminology. For the purposes of our present discussion, we can still represent these philosophers as accepting a thesis of perceptual mediation in the sense defined, and as taking the internal objects of awareness to be the things whose occurrence plays the relevant mediating role.

Now when the direct realist claims that, at the point of Φ-terminality, our perceptual contact with the physical world is perceptually direct, it is precisely this form of the representative theory that he is wanting to exclude. He is insisting that the perceiving of the relevant physical item is not subject to perceptual mediation, and the point of this insistence is to rule out this traditional empiricist alternative, which takes the perceiving of the item to be mediated by the occurrence of an internal object of awareness, which represents it. This gives us, then, our first way of construing the direct realist position and the issue between it and the representative theory. Direct realism, thus construed, claims that Φ-terminal perceiving is always perceptually direct; and, in appropriate opposition, the representative theory claims that such perceiving is always perceptually mediated by the perceiving, or sensing, of something in the mind—something whose existence is necessarily confined to the context of the subject's own awareness.

For the sake of precision, there is something which I should now add, though its significance will only become apparent later. As I have so far characterized it, the relevant version of the representative theory takes the immediate objects of perception to be entities which are internal to the subject's awareness; and I have taken this to mean that, on each perceptual occasion, the immediate object is something which has, and can have, no existence outside the context of the particular episode of awareness directed on to it. However, although this is certainly the standard way in which this form of the representative theory is held, it is not the only possibility. For it is also possible for the theorist to construe the relevant objects as a special category of universals—as things which are capable of occurring to different subjects and on different occasions. And, on this construal, we would have to say that what is tied to the context of the subject's awareness on a given occasion is not the existence of the relevant

end p.9

 

object, but only a certain occurrence, or realization, of it. The availability of this construal will become crucial at a certain stage in our investigation. But, at present, it will only tend to undermine our understanding of the situation if we try to take it into account. And so, until the critical stage is reached, I shall simply put the point on one side, and continue to think of the relevant theory in its standard form—with the objects tied, in their existence, to particular subjects at particular times.

III

The issue we have just identified is concerned with the scope of perceptual mediation: it is the issue of whether, at the point where perceptual mediation terminates in the physical domain, our perceptual contact with the physical world is perceptually direct, or is rather subject to a further stage of such mediation, involving the perceiving, or sensing, of items in the mind. But perceptual mediation is not the only form of mediation which could be thought of as relevant. For philosophers who agree that Φ-terminal perceiving is not subject to any further form of perceptual mediation, may still differ over whether it is subject to mediation in a broader sense, and this difference of view can also be represented as an issue between a kind of direct realism and a kind of representative theory. Let me explain.

Suppose we have a subject S who perceives a physical item x at a time t. Then let us say that S's perceiving of x at t is psychologically mediated by his being in Σ if and only if

(1)	Σ is a psychological state;
(2)	Σ is not, in itself, x-perceptive (i.e. being in Σ does not, on its own, logically suffice to put one in perceptual contact with x);
(3)	S's perceiving of x at t breaks down into his being in Σ at t and certain additional facts; and
(4)	these additional facts do not involve anything further about S's psychological condition at t (anything over and above what is already covered by S's being in Σ). In other words, in combining with the fact of S's being in Σ, they do not add any further psychological facts, about S at t, to the constitutive base.

end p.10

 

In a similar vein, let us say that S's perceiving of x is psychologically mediated (tout court) if and only if there is some state Σ such that it is psychologically mediated by his being in Σ, and that S's perceiving of x is psychologically direct if and only if it is not psychologically mediated. I should mention here that when I speak of a 'state', I always mean a type-state, not a token-state. In other words, a state, in my terminology, is a sort of universal—something which, at least in typical cases, is capable of being realized in different things and at different times.² I should also stress that states, as I construe them, can be relational, so that we can speak of the psychological state of seeing a particular apple or of being in love with a particular person.

Now it is with respect to psychological mediation that the further issue, between a kind of direct realism and a kind of representative theory, is to be defined. Suppose S Φ-terminally perceives a physical item x. If we claim that S's perceiving of x is perceptually mediated, we shall also have to accept that it is, eo ipso, psychologically mediated too. For if the perceiving is perceptually mediated, then it must break down into the occurrence of an internal object of awareness and certain additional facts of the relevantly restricted kind; and this will mean that there is a psychological state Σ, which is either simply the state involved in the awareness of the internal item (if the additional facts do not include additional psychological factors), or is this state together with some enrichment (if the additional facts do include additional psychological factors), such that S's perceiving of x is psychologically mediated by his being in Σ. But crucially, this point does not hold in reverse. Thus if we claim that S's perceiving of x is psychologically mediated, this does not commit us to saying that it is, eo ipso, perceptually mediated. For we might want to say that the perceiving is psychologically mediated by S's being in a state which is not itself perceptual—a state which does not involve the occurrence of a perceptual object in the mind. For example, we might want to say that what psychologically mediates the perceiving of x is not the occurrence of some internal object of awareness (or that, together with certain other psychological factors), but the subject's acquiring of a certain belief, or complex of beliefs, about the current state of his environment.

It is this which creates the possibility of an alternative interpretation of direct realism and its conflict with the representative theory. The direct

² I say 'at least in typical cases', because it is possible for a state—a type-state—to be defined in a way which restricts its capacity for realization to just one thing, or to just one time, or, indeed, to just one thing at one time. For example, the state of being taller both than anyone else at any time and than anyone at any other time can at most be realized by one person at one time.

end p.11

 

realist claims that our perceptual access to the physical world is direct, and, given that there are uncontroversial cases of perceptual mediation within the physical domain, this boils down to the claim that Φ-terminal perceiving is direct. But, now, instead of taking this claim to mean that Φ-terminal perceiving is always perceptually direct—not subject to perceptual mediation—we could take it to mean, more strongly, that such perceiving is always psychologically direct—that it is not subject to psychological mediation in any form. And correspondingly, instead of taking the representative theory to be claiming, restrictively, that Φ-terminal perceiving is always perceptually mediated—a mediation which involves the occurrence of a perceptual object in the mind—we could take it to be claiming, more flexibly, that it is always psychologically mediated in some form—without commitment to whether the psychological state involved is itself (wholly or partly) perceptual. It might be thought that, in formulating this more flexible position, I should at least have represented it as insisting that Φ-terminal perceiving is psychologically mediated in some representational form; for the claim can hardly count as a version of the representative theory unless it restricts the relevant forms of psychological mediation to ones which involve representation. But this is not, I think, a point of real substance. If the realization of the psychological state succeeds in mediating perceptual contact with the physical item in question, it is bound to be something that we can think of as in some way representing (in some way serving to convey the presence of) that item to the subject involved. At least, this is bound to be so once we have agreed that psychological representation does not have to take the perceptual—as it were pictorial—form of the classic empiricist version of the theory. In any case, however we care to label it, it is the flexible position as I have actually formulated it—the position which simply claims that Φ-terminal perceiving is psychologically mediated—that will be relevant to our concerns.

IV

The traditional issue between direct realism and the representative theory has now turned into—or turned out in reality to be—two issues. Thus, on the one hand, there is the issue between a weaker, more modest version of direct realism, which claims merely that Φ-terminal perceiving is perceptually direct (not subject to perceptual mediation), and a narrower, restrictive version of the representative theory, which claims, in opposition, that such perceiving is perceptually mediated (psychologically mediated in a way which involves the occurrence of a perceptual object in the mind). Let us

end p.12

 

label these positions weak direct realism (WDR) and the narrow representative theory (NRT). On the other hand, there is the issue between a stronger, full-blooded version of direct realism, which claims that Φ-terminal perceiving is psychologically direct (not subject to psychological mediation of any kind), and a broader, more flexible version of the representative theory, which claims, in opposition, that such perceiving is psychologically mediated in some way (without commitment to whether the psychological state involved is itself perceptual). Let us label these positions strong direct realism (SDR) and the broad representative theory (BRT).

An acceptance of the stronger version of direct realism (SDR) would automatically commit us to an acceptance of the weaker version (WDR), and an acceptance of the narrower version of the representative theory (NRT) would automatically commit us to an acceptance of the broader version (BRT). And so, a fortiori, SDR (standing in opposition to BRT) and NRT (standing in opposition to WDR) are themselves incompatible. But, crucially, WDR and BRT—the weaker version of direct realism and the broader version of the representative theory—can be consistently combined, since there is the option of saying that Φ-terminal perceiving is psychologically mediated, but in a non-perceptual way. It is because of this that our discernment of the different versions of direct realism and the representative theory counts as a reshaping of the traditional issue, rather than as just an elaboration of it. If each version of each of the traditional positions turned out to be incompatible with each version of the other, the traditional issue would remain as something well defined, and the discovery of the different versions would just be the discovery that the traditional positions—on each side of a clear-cut divide—were capable of further subdivision. The new issues to emerge would not be between direct realism and the representative theory, but within them. As it is, the traditional issue has shown itself to be ill defined, there being no clear-cut divide between the positions which feature in it; and, in its place, we have to put the two properly defined issues which we have just identified, each with its own precise conception of how the distinction between directness and mediation is to be drawn.

In our quest for an understanding of the nature of perception, each of these issues is crucial to our concern, and will need to be addressed at some point. But the first thing we have to decide is the order in which to take them. Either order would be procedurally possible. Thus, on the one hand, we could start by considering the conflict between the weaker version of direct realism (WDR) and the narrower version of the representative theory (NRT). If NRT emerged the winner, this would automatically settle

end p.13

 

the other issue in favour of BRT, since NRT entails BRT. If WDR emerged the winner, we could then turn our attention to the consequential issue of whether we should hold the direct realist position in its full-blooded (SDR) form or combine the weak direct realist view with the acceptance of some non-perceptual form of psychological mediation (in accordance with BRT). On the other hand, we could start with the conflict between the stronger version of direct realism (SDR) and the broader version of the representative theory (BRT). If SDR emerged the winner, this would automatically settle the other issue in favour of WDR, since SDR entails WDR. If BRT emerged the winner, we could then turn to the question of the precise form of psychological mediation involved, and, in particular, of whether we should hold the representative theory in its narrow, perceptual-mediational, form (NRT) or combine it with the weak version of direct realism (WDR).

Although both these investigative procedures are available, and would end up covering the same ground, the second is clearly preferable. This is because the range of positions left open by BRT forms a natural genus, whereas those left open by WDR do not. Thus the BRT-positions all have in common the positive feature that they take Φ-terminal perceiving to be psychologically mediated in some way—the difference between them only concerning the nature of the psychological state involved; whereas what the WDR-positions have in common is the negative feature that they exclude perceptual mediation at the point of Φ-terminality—a feature which could take the very different forms of a full-blooded direct realist view (denying psychological mediation altogether) or a type of representative account which postulates psychological mediation of a non-perceptual kind. Since it would be awkward having to deal with such contrasting positions at the second stage of the investigation, it is more natural and more rational to begin by considering the issue between SDR and BRT—the issue of whether there is psychological mediation at all—and then, should it arise, turn to the consequential issue of the form of mediation involved. This, at any rate, is the procedure I shall follow.

3 SDR and BRT

I

The issue on which we shall be focusing, then, in the first phase of our investigation, is that between the strong, full-blooded version of direct

end p.14

 

realism (SDR), which claims that Φ-terminal perceiving is always psychologically direct, and the broad, flexible version of the representative theory (BRT), which claims that such perceiving is always psychologically mediated. BRT could equally be formulated as the claim that all physical-item perceiving (whether Φ-terminal or not) is psychologically mediated. For any psychological mediation which applies at the point of Φ-terminality is bound to apply to the perceiving of any further (perceptually more remote) physical item which the Φ-terminal perceiving perceptually mediates (though if the perceptual mediation involves some additional psychological factor, the relevant Σ will have to be suitably expanded). The only point in restricting the formulation of BRT to the Φ-terminal case is that it is exclusively in relation to this case that the conflict with SDR arises.

This conflict is over the question of whether, at the point of Φ-terminality, the perceptual relationship between the subject and the relevant physical item decomposes in a certain way. In claiming that it is psychologically mediated, BRT is claiming that this relationship breaks down into (is constituted by the combination of) two components, the one consisting in the subject's being in some more fundamental psychological state—a state which is not, in itself, perceptive of the relevant physical item—the other comprising certain additional facts, but ones which do not involve anything further about the subject's psychological condition at the relevant time. In claiming that the perceptual relationship is psychologically direct, SDR is claiming that it does not decompose in this way—that it does not break down into these two types of component in the way that BRT envisages. So the two positions are diametrically opposed, one asserting the universal obtaining of the relevant kind of decomposition, the other asserting its universal absence.

Although, thus defined, SDR only formally excludes this particular form of decomposition—the decomposition definitive of psychological mediation—an acceptance of SDR would, in effect, involve accepting something stronger. Thus it would, in effect, involve accepting that, at the psychological level of description, Φ-terminal perceiving does not decompose in any way at all—that where someone Φ-terminally perceives a physical item, his perceptual relationship with it does not break down into any plurality of facts which are wholly or partly concerned with his psychological condition. Indeed, it would in effect involve accepting that, at the psychological level, the perceptual relationship is something fundamental—something which is not subject to constitution of any form. The only qualification here—though the significance of this will only emerge later—is that the SDR-theorist does have the option of saying that the bare fact of

end p.15