INTRODUCTION

 

Think of the range of visual images available within our culture and the roles each kind plays in our lives. We attend to ourselves by looking at our mirror images; we aid our failing memories by gazing at photographs; we admire how a painting evokes the essence of a landscape or a character; we weep as we stare upon a cinematic image glowing from a silver screen; we entertain and inform ourselves by sitting in front of a television; we admire a quick sketch with captures a person’s looks; and, as Wordsworth remarked, we lie upon our couches, in vacant or in pensive mood, allowing our mental images to flash upon our inward eyes, bringing bliss to our moments of solitude. The visual image it seems is something we could scarcely be without. We have an endless fascination and preoccupation with visual imagery. It is ubiquitous. Indeed, it would be hard to see how our culture could survive without it. The rich variety of visual imagery is, without doubt, the most familiar of phenomena. Yet it is a profound and frequently encountered lesson of philosophy - and perhaps one of which we are all constantly needful - that familiarity is not the same as understanding.

 

My aim in this book is to isolate and examine contending theses, some my own, some  developments of the ideas of others, about the fundamental nature of various forms of visual imagery and how these natures require that we categorise visual imagery into fundamental types. Thus, it is a work primarily of metaphysics and ontology. Despite the difficulties in arriving at a settled view on these and related matters, I will nevertheless heartily attack certain theses and equally vigorously defend others.

 

How does our material carve up? First there are issues concerning the origination of visual imagery. There is a large group of visual images which depend crucially on a high degree of non-intentional, largely mechanical, causation. Cameras, scopes and mirrors typically produce their images in ways which depend purely on the laws of  nature. Crudely, we could say, all you have to do is set up the apparatus, and  natural causation does the rest. Painting and drawing, on the other hand depend crucially on the artist’s intention for the production of the image; one cannot just set up the apparatus and expect nature to produce the images for you. Mental imagery is perhaps something of a half-way house. Some mental imagery is intentional in almost the way painting is. That is, we build it up deliberately to create the image we seek. Much of it, however, just comes unbidden and by means that depend more purely on natural causation.  And even mental imagery which is deliberately constructed relies heavily on the processes of natural causation. It is my contention that images produced in this way have certain properties which require special philosophical attention in addition to the problems of depiction and representation which also feature in our understanding of painting. 

 

But why does imagery which depends so largely on the processes of  natural causation deserve special attention? To see this we need to understand visual imagery as varying in its (most fundamental) nature with respect to two important qualities. First, I will contend, imagery may be either iconophilic or iconophobic.[1]  This means that the imagery under consideration either does or does not involve the seeing of an actual, literal image. Many philosophers would say, for instance that mental imagery must be explained in iconophobic terms. The reason for this, as we shall come to see, is that positing literal images in the head (or mind) of the imager gives rise to a whole raft of philosophical problems. We also ask whether other forms of imagery are iconophobic. Photography certainly isn’t but the answer is less clear when it comes to images produced by mirrors and scopes. This is an issue to be discussed.

Secondly, come issues connected with the notions of transparency and opacity. Some images seem undoubtedly opaque. This means that in looking at the image, the image is all we literally see. Painting, drawing and computer generated imagery seem firmly to fall into this category. In seeing a painting of a rose, for instance, I see just that - a painting.  I may see what is in the painting as a rose but the painting is all I literally see. It can be argued, however, that some forms of imagery have the quality of transparency. This means that in looking at the image, I see more than the image, I see through the image itself to the thing which the image is an image of. Mirrors and scopes (at least ordinary kinds) may be said to produce images which posess this quality. If I see a rose in a mirror or telescope, for instance, then if I see the image, I also see the rose. The image is, we might say, a means by which I see the rose. To say that an image of x is transparent means, in other words, that the image provides one with (a degree of) perceptual access to x. That images produced by scopes and mirrors are transparent may seem as indubitable as the thought that paintings are opaque, but we shall see that the issue is not quite that simple. Arguably, however, even photography produces transparent images, and this will be a position I will, with certain modifications, defend. In addition to the questions already posed about painting then, further questions will be whether instruments such as mirrors, scopes and cameras extend rather than merely aid our ordinary perceptual access to the world, and if so, in what ways. 

 

We also ask how we are to construe mental imagery. In particular, we focus on the question of  how mental imagery may be said to be relational. Most imagery is quite uncontentiously relational. That is to say, any analysis of the experience of such imagery must admit the existence of (at least) two things - the imager and the image (or the imager and the imaged object). We would say that the imager is perceptually related to one or both of these items. It is commonly contended, however, that mental imagery is unique in being non-relational. There is an imager who is enjoying an image-type experience but there is no item to which the imager is perceptually or quasi-perceptually related. I will, however, argue that there is good reason to think that mental imagery must involve a perceptual relation between the imager and some other item, that it must, in other words, be relational.

 

In summary, then , the three central issues of this book will be those of transparency versus opacity, iconophilia versus iconophobia and (in connection with mental imagery alone) relationality versus non-relationality. These issues are, I hope to show, clearly metaphyisical in nature. This means that there is simply no hope of any empirical or scientific determination of them. It also raises the question as to how we may hope to resolve such issues. How, for example, can we decide whether a candidate form of imagery is transparent or opaque? One clear way to proceed would be to start off with some means by which we are undoubtedly judged to be perceptually in touch with the world and then compare the target form of imagery with that. Ordinary vision is the obvious candidate, but there are complications. A sceptic, for instance, may worry whether even ordinary vision puts us in touch with the world at all. I take sceptical challenges of this sort seriously, though for the purposes of the present book, I will simply set such concerns to one side. More pertinently, even non-sceptical Representationalists contend that in perception what we directly see are the contents of our own heads or minds, and that we see the world outside only indirectly, or perhaps infer its existence from our inner experiences. We will need to examine this issue later on (when we consider the philosophical arguments for Iconophilia in respect mental imagery). We will, however, start with the unargued assumption that ordinary perception puts us in touch with the world if anything does. Our strategy will then be to compare various forms of imagery with this feature of perception. Then, to the extent that we find that the imagery in question shares this capacity with ordinary perception, we will declare it to be (to that extent) transparent. Note that this means that transparency is going to be a matter of degree, something which will prove to be important later on. The other two qualities we have isolated cannot be matters of degree. A form of imagery may be either iconophilic or iconophobic but not in between. The same goes for the relational / non-relational distinction. (A form of imagery may contain both iconophilic and iconophobic elements or parts. It is entirely probable that mental imagery does, but to concede the probability of a mixture of different parts is not to concede the possibility an in-between position.) As to these other two issues – those involving the iconophilia / iconophobia debate and the relationality / non-relationality debate – our strategy towards resolution will follow the usual philosophical method of testing the arguments put forward in favour and against the various positions for validity and soundness.

 

Note also that our three distinctions are not entirely logically independent. At least one combination seems ruled out conceptually and, thus, cannot feature in any theory of imagery. The combination of iconophilia and non-relationality is ruled out on the grounds that if there are literal images involved in any form of imagery, then it is presumably these that we see. On the face of things it may seem as if  the combination of iconophobia and transparency is also ruled out, for if there is no literal image, how can it be transparent? But this would be too hasty a move. The experience itself may still be transparent in the sense that it may be a way of seeing things beyond. This, in fact, will be the basis of the World-Directed Iconophobia thesis discussed in Chapter 19.

 

The plan is as follows. First, I discuss transparent imagery. Mirrors and scopes come first (Chapter 1). In Chapter 2, I make out the case for photographic transparency and in, Chapter 3, I defend the thesis against some common complaints. In Chapter 4, I extend the application of transparency to motion pictures. The rest of the book consists of an examination of mental imagery. 

 

Although I could have kept the second part shorter simply by showing how transparency can be applied to mental imagery, the requirement for a more extensive treatment arises from the extraordinary richness and complexity of the issues raised by the topic -  issues which span various parts of the philosophy of mind, perception and even language. A proper understanding of mental imagery seems impossible, therefore, without taking such a wide sweep, but since width is useless without the appropriate depth, the domination of the book by the discussion of mental imagery was inevitable. 

 

I begin my treatment of mental imagery, in Chapter 5, by outlining the philosophical problems to which the phenomenon gives rise and in particular make much clearer what is meant by the distinction between relationality and non-relationality (a distinction which, as I have intimated, can only apply to mental imagery). The next two chapters deal with  some well known empirical findings (Chapter 6) and philosophical arguments (Chapter 7) which may seem to favour iconophilia - the thesis that mental imagery consists of literal images in the head (or mind) of the imager. My aim here is largely negative; I will show how appeal to such matters is quite indecisive. 

 

Since iconophilia implies relationality (if literal mental images exist, it is presumably these that we, in some way, apprehend) the next part of the book takes up the issue of relationality. Central to this is the notion of phenomenality - how we are to account for the occurrence in visual experience of certain phenomenal properties (e.g. colour). This notion is crucial to arriving at a proper understanding of mental imagery and, as we shall see, places certain constraints on  possible theoretical perspectives on the phenomenon. I call this the ‘constraint of phenomenal adequacy’. It too, therefore,  requires extensive treatment. In an attempt to be as clear as possible about this issue, I consider it in connection with perception (Chapter 8), belief (Chapter 9) and imagery itself (Chapter 10). This chapter, in fact, contains some arguments (based on the two earlier chapters) to show that when I image, say, grass, something green is phenomenally before my mind. In other words, I will argue that the phenomenon is irreducibly relational. This is the first of a number of constraints that we will find must be placed upon any adequate theory of imagery.

 

In the next three chapters I go on to consider three other putative constraints which, it is often thought,  should be met by any decent theory of imagery. The first demands that any theory should meet the constraint of phenomenological adequacy. That is, it should be clear as to just what kind of phenomenological properties mental imaging has. This, as we come to see, covers a wider range of issues than the ‘constraint of phenomenality’ mentioned above. Here we will ask just what mentally imaging something is like. In particular we shall enquire whether it is like seeing something, and, if so, in what ways and to what extent. The next putative constraint is the constraint of non-incorrigibility. That is to say, no theory of mental imaging should necessitate the positing of incorrigibly known objects or mental particulars – items that can be known by the imager in ways that lie beyond the capacity of anyone to correct. Finally comes what I shall call the constraint of non-privacy. This demands that no theory of mental imaging should necessitate the positing of items which are radically mentally private - private, that is, in terms of the widely accepted arguments of Wittgenstein. 

 

I say these are putative constraints since it is not clear, in advance of discussion, just what these constraints can impose upon possible theoretical perspectives, although a common charge against iconophilia is that fails (indeed, must fail) to meet all of these constraints. For example, it is often thought that mental images have (or can have) a built-in indeterminacy and that since lietral images cannot have the required form of indeterminacy, mental imagery cannot consist in the apprehension of literal images. (This is one aspect of the phenomenological constraint.) These issues are well enough known to philosophers, but familiarity is not the same as clarity and, whatever the outcome, it is necessary to get clear as to just what sorts of properties mentally imaging does in fact have and just what constraints can be imposed. That is the job of Chapters 11, 12 and 13.

 

Getting clear just what constraints are imposed (if any) clears the ground for our discussion of iconophilia. Three possible kinds of iconophilic theory are considered in the following three chapters. Chapter 14 deals with what I call Non-Physical Iconophilia, the view that mental images are non-physical particulars. Chapter 15 deals with something I call Information Structure Theory. This is the view that mental images are physical information structures in the imager’s brain. Both of these theses are rejected, the first on fairly familiar metaphysical grounds, the second on the grounds that it both lacks clarity and fails to satisfy the constraint of phenomenal adequacy. Finally, in this section, Chapter 16 puts forward a new form of iconophilia - something I term  Phenomenal-Structure Theory. This holds that mental images are both physical and phenomenal structures in the brains of imaging subjects. Though I eventually come to reject this theory too, I do not think it can be decisively refuted and I offer it as the iconophile’s last and best hope. My intention here is to show that iconophilia is at least no philosophical dead duck (as is so frequently assumed).

 

Many (probably most) philosophers, however, have been sufficiently convinced by the difficulties facing any kind of Iconophilia to seek out alternative strategies for dealing with mental imagery. The next part of the book deals then with various iconophobic approaches to mental imagery. These are ways of accounting for mental imagery without positing literal images in the head or mind of the imager. My aim will be first of all to show how all the popular varieties of Iconophobia fail to satisfy the constraint of phenomenal adequacy. I will then present two possible iconophobic theories which, I will argue, do meet this constraint.

 

One can, I believe, isolate two broad popular iconophobic strategies. The first of them is particularly associated with three philosophers - Dennett, Ryle and Wittgenstein. Dennett, while claiming in places to be open minded about the nature of imagery, seems, I will argue, to favour an interpetationalist view - images are ways of understanding an imager’s behaviour and are purely notional, predictive devices. Ryle, on the other hand, considered mental imagery to be a form of pretending to see, while Wittgenstein laid stress on the need for outward criteria. Thus, we have here three views (if not altogether settled theories) which, while they are very different in content, are united by their emphasis on the role played by behaviour and context in our criteria for ascribing imagery. I will thus call this kind of strategy Quasi-Behaviourist Iconophobia. (Though it would, of course, be wrong to characterise Dennett, Ryle or Wittgenstein strictly as behaviourists.) This I discuss in Chapter 17.

 

In contrast, the second kind of strategy lays stress upon what is going on inside us when we image: briefly, the thesis is that whatever is happening inside my brain when I image is like what happens inside my brain when I perceive. While not associated with any particular philosopher, it is strongly favoured by those of a physicalist persuasion who wish to find a way of accommodating the phenomenal dimension of imagery within an iconophobic framework. I thus call this strategy Central State Iconophobia. This strategy I will examine in Chapter 18.

 

A further aim in this part of the book, however, is to develop  two very different forms of Iconophobic theory. The first has its origins in Sartre, though, I will argue, he failed properly to develop his own insight. Briefly, the idea is that mental imagery is just a delayed and often transformed perception of previously perceived states of or objects in the world. In the basic case, when I image x, I am quite simply having a late perception of x, a re-experience of x, a perception that is mediated by a longer causal chain than is involved when I literally perceive x before my eyes. This I call World Directed Iconophobia. This approach is a theoretical reversal of Indirect Realism or Representationalism. Starting from the phenomenal similarity of imaging and perception, the Representationalist construes them both as forms of imagery; whereas the World-Directed Iconophobe argues that they both be construed as forms of perception of the world. This is discussed in Chapter 19. Of course, not all cases of imagery are like the basic case described above and some obvious problems arise when one considers cases of imaging objects that have not been previously seen. Such difficulties may, indeed, inspire sheer incredulity at the whole notion of World-Directed Iconophobia, but I think that the difficulties can be met and the theory vigorously defended. World-Directed Iconophobia takes seriously the thesis that instantiations of colour are always instantiations of a property of some object. Since, as our arguments will show, we just do apprehend colour as we image, the thesis has to find suitable bearers of the colour we apprehend. Indeed, all our favoured theses so far rest upon the unargued assumption that the appearance of colour is a property and thus requires a bearer. I therefore come finally in this section to question this assumption and consider a new form of iconophobia - something  I call Pure Event Iconophobia – which explicitly rejects it. Pure Event Iconophobia holds that our perception of colour in imagery is to be accounted for in terms of its being a pure event rather than in terms of our perceiving or somehow apprehending objects which possess colour. This is the subject of Chapter 20. I finish (Chapter 21) with some concluding remarks about mental imagery. 

 

It may here be queried why I do not include a discussion on the nature of the opaque image. Remember that the most fundamental distinction to be observed when considering the nature of visual imagery centres upon the way the image is produced. At rock bottom there are just two ways of producing visual images - you either let nature do the job or do it yourself. Images seen as reflections in mirrors or as the products of lenses (including photography, cinema and television) fall firmly under the former category; images produced by painting, drawing and their ilk, fall equally decisively under the latter. In saying this I do not mean to deny the obvious truth that photography involves a number of human intentions and purposes. Rather, I mean that once the apparatus (camera, darkroom equipment, and so on) has been set up, photography depends upon purely natural processes (optics, chemistry, and so on). Photography and painting, though equally man-made, differ in this respect. Painting crucially depends on human intention and photography on nature’s automated process of cause and effect. 

 

That this is a real difference in the mode of image production between photography and painting may be accepted, but it may still be queried why we should think the difference crucial. The full answer to this question will take up many chapters but here we can ask the following two important questions. What sort of thing is a painting?[2] And what can we be said to see when we look at a painting? Paintings are, of course, iconophilic: to look at a painting is to look at a literal image. Let us leave that as obvious. Paintings are also opaque: when we look at a painting of a man, we literally see only the painting. 

 

Painting also exhibits, as we have noted, a certain kind of intentional dependency: what you literally see hangs entirely upon what the artist has in mind. I do not mean by this to deny that a painting may be interpreted in a wide variety of ways, some of which may not have been intended by the artist. Rather, I mean simply that the arrangement of coloured patches on the canvas depends on the intentions of the artist: there is a patch of green here and a streak of red there because that is what the artist wanted, or, if accidental, what the artist thought would do. Insofar as we can say that we see, say, a man in a painting at all, we must understand this as meaning that we see a man represented in the painting: we may say that we see not a man but, rather, a man in the painting. For this reason we may say that painting affords no degree (or kind) of perceptual access to the man himself. Indeed, it is not even necessary that the man exists. This contrasts, I contend, with (at any rate, straightforward) photography, in which the arrangement of colours depends on the features of the scene being photographed. In a photograph, there is a patch of green here and a streak of red there because nature (including the  causal processes of photography) put them there. One way of characterising this difference would be to say that a painted image of a tree is put there by an artist, whereas the photographic image of a tree is put there by (among other things) the  tree itself. As we shall come to understand, this feature of photography gives ground for believing that photographs give us some (albeit very limited) degree of perceptual access to their objects. If we see a photograph of a man we quite literally see not only the image of the man but also (though indirectly) the man himself  - or so I shall argue. 

 

Painting, then, is (presumably quite uncontroversially) iconophilic (and thus relational) and opaque, so it does not seem to raise any important metaphysical question relating to our three core issues. To be sure, there are some important philosophical questions that relate to painting. We may ask, for example, how we can be said to see anything at all in a painting. If when we look at a painting what we literally see are patches of colour on a canvas, just how can we see anything in it? One obvious answer is that the painting is, in some sense, a copy of  its subject. But in what sense?  One popular answer has been to say that pictures copy the reality they depict, but there are many problems with this simplistic view. Fascinating as these questions are, however, they are not our questions. They belong to the areas of Theory of Depiction and Representation rather than metaphysics – and we restrict ourselves to the latter.[3] Discussion of opacity and painting, therefore, will only occur where we contrast it with transparency of photography.

​

[1] I take these terms from Dennett, D Brainstorms. Dennett used them in a somewhat different way and in connection only with mental imagery. I use the terms in my own way and extend them to other forms of imagery.

[2] For our purposes I include drawings, etchings, pastels, indeed, any piece of art produced by the hands of an artist. I would also include computer art (of which more later), especially computer art which does not originate photographically.

[3] For a recent and thorough survey of these areas and a vigorous defence of a form of the Copy Thesis of depiction, see Robert Hopkins,  Picture, Image and Experience, 1998.