The Unity of Pictorial Experience

1. Modes of Visual Awareness

The first thing to get clear on is the notion of a mode of visual awareness. By a mode of visual awareness, I mean a way in which an object or scene can be visually experienced. To borrow from Sartre, a mode of visual awareness is a ‘way in which [an object or scene] appears to [visual] consciousness’ (Sartre 2010/1940, 7). In what follows, I will reserve the word ‘perception’ and its cognates for a very specific mode of visual awareness. I’ll say that our visual awareness of something is ‘perceptual’ only if it is experienced as present to the visual senses. It does not matter if the object is actually present to the visual senses; as long as it is experienced as such, I will say that it is perceived. Thus, hallucinating a castle involves experiencing one in the perceptual mode, whereas visualizing (i.e., visually imagining) a castle does not. Visualizing a castle is a way of being visually aware one, but it is not an instance of the specific form of visual awareness that I am calling ‘perception’: when we visualize a castle, we do not experience it as present to the visual senses. In short, perceiving a castle (or hallucinating one) involves one mode of visual awareness, and visualizing a castle involves another.

As these examples make clear, the very same thing can appear in different visual modes; characterizing a visual experience means specifying both its content and the mode of visual awareness in which it appears. As Sartre says, ‘whether I perceive or imagine [the straw-bottomed chair on which I sit], the object of my perception and that of my [visual imagining] are identical: it is that straw-bottomed chair on which I sit’ (loc. cit.). Where Sartre speaks of the ‘object’ of a visual experience, I will speak of its ‘content’. By the ‘content’ of a visual experience, I mean the sort of thing that can be specified by completing a phrase, such as ‘S perceives…’ or ‘S sees…in…’, in which a mode (or modes) of visual awareness is (or are) specified. The content of a visual experience such as that specified by ‘S perceives…’ could be any of the following: a woman, my sister Ana, the colour red, that this rose is red, that something is red, that Jack is next to Jill, and so on.

Experiences of seeing-in are special because they are twofold: they involve two modes of visual awareness of two different objects.² We can put this by saying that the form of the sentence ‘S sees O in P’ is different from the form of the sentence ‘S perceives a cat in a hat’, or ‘S perceives a castle through the window’. Whereas the latter two sentences mention one mode of visual awareness (‘S perceives…’), the former mentions two (‘S sees…in…’). It also has two gaps for content. ‘S sees O in P’ mentions the perceptual mode of visual awareness that we have of P, and the non-perceptual mode of visual awareness that we have of O. I should flag that some people have thought that O is also experienced in the perceptual mode, just as P is. They do not think that O is actually present to the visual senses, but they do think it appears to us as though it is. This is the view of the so-called ‘transparency theorists’ (Kulvicki 2009; Newall 2015). According to them, when we see O in P, we perceive O through P. (Since transparency theorists also think that P is nonetheless perceived as an opaque array of ink-marks, their view introduces an element of contradiction into pictorial experience (Hopkins 2012).) The cost of this view is high. Since we do not believe that O is present to the visual senses, any view on which O is nonetheless experienced in the perceptual mode is committed to holding that our experience calls for correction at the level of belief. This illusionism is hard to swallow. Holding firm to our belief that O is absent does not appear to require anything like the kind of cognitive effort that it takes to believe that the Müller-Lyer lines are equivalent in length.

For the purposes of this paper, I am going to assume that these views are false. When we see a castle in a picture, we are visually aware of a castle, but we do not experience it in the perceptual mode. In holding this, we do not have to think that seeing O in P therefore involves visualizing O. (Thus, we need not agree with Malcolm Budd that ‘the only alternative to an experience as of face-to-face [perception] is the second principal form of experiential visual awareness – visualizing what is not present to the eyes’ (Budd 2008, 197). Instead, we can hold that seeing O in P involves being aware of O in a sui generis non-perceptual mode of visual awareness that is not subsumed under a more specific genus than ‘mode of visual awareness’, but which lies alongside perception, visualization, and so on, in a taxonomy of the different modes of visual awareness (see Figure 1). This is what I think we should hold, but anyone who disagrees with me will not be disbarred from accepting the paper’s main contentions below (see fn. ⁶).

Figure 1

Because seeing-in is twofold, there is no such thing as the mode of visual awareness that it involves; two modes of visual awareness are operative when we see O in P.³ Although introducing terms of art always runs the risk of making things less intelligible instead of more, it will be helpful to give these modes of awareness names. I will call the perceptual mode of visual awareness that we have of a marked surface (any marked surface) when we see an object or scene (any object or scene) in it ‘p-vision’, and I will call the non-perceptual mode of visual awareness that we have of an object or scene (any object or scene) when we see it in a marked surface (any marked surface) ‘o-vision’. It is important to emphasize that these terms do not introduce technical notions. Anyone who accepts that seeing-in is twofold, i.e., that seeing O in P involves being visually aware of both O and P, believes in o-vision and p-vision. This paper could be written without these expressions, but the alternative, which is to continually use the descriptions that fix their referents, would be unbearably cumbersome. For ease of exposition, I’ll also say that S ‘p-sees’ something when S is visually aware of it in the p-vision mode, and I’ll say that S ‘o-sees’ something when S is visually aware of it in the o-vision mode.

2. Reconciling Wollheim’s Two Claims

Seeing-in is twofold, but it is also a single experience. As Wollheim writes:

the two things that happen when [I see-in] are, it must be stressed, two aspects of a single experience that I have, and the two aspects are distinguishable but also inseparable. They are two aspects of a single experience, they are not two experiences. They are neither two separate simultaneous experiences, which I somehow hold in my head at once, nor two separate alternating experiences, between which I oscillate. (Wollheim 1987, 46)

Various people have said similar things. For example, here is Wittgenstein:

although the expression that seeing a drawing as a face is not merely seeing strokes seems to point to some kind of addition of experiences, we certainly should not say that when we see the drawing as a face we also have an experience of seeing it as mere strokes and some other experience besides. And this becomes clearer when we imagine that someone said that seeing [the Necker cube] as a cube consisted in seeing it as a plane plus having an experience of depth. (Wittgenstein 1969/1958, 169)

Wollheim’s statement has caused some confusion. For instance, Kendall Walton writes: ‘Seeing-in is an experience characterized by what [Wollheim] calls “twofoldness”: one sees the marked picture surface, and one sees the subject of the picture. These are not two independent experiences, but two aspects of a single one. It is hard to know what this means, and Wollheim offers little explanation’ (Walton 2002, 33). Similarly, John Hyman has complained that, since Wollheim ‘does not explain how he believes experiences should be counted, it is difficult to assess [his claim]’ that ‘being visually aware of the surface and discerning something [in it] are two aspects of a single experience’ (Hyman 2006, fn. 6). It is true that Wollheim was not as explicit as we might have wanted on this point, but we can make some progress by elaborating on it as follows:

single experience: E is a single experience if, and only if, E presents to the experiencing subject as being a single experience.

This definition is not completely satisfactory, insofar as it itself deploys the idea of a single experience, but it is what I propose to work with for the time being. When we see O in P, it does not appear to us as though we are undergoing two separate experiences simultaneously. We are not aware of having two completely independent experiences, as we are when we perceive A while simultaneously visualizing B (Hopkins 2012; 1998, 17), and we are not aware of having two interdependent experiences, either. On the contrary, seeing O in P presents to us as a single, unified whole. As I said above, part of the point of the discussion to come is to deliver a more precise understanding of this claim.

We can get into a position of doubting this feature of seeing-in because some pictures allow for switching rapidly between different sorts of experience of their surfaces. For instance, a highly stylized representational painting might lend itself to seeing parts of its surface as meaningless marks in one moment, before seeing the painting’s (whole) content in another. Because this transition can happen quickly, and because the first experience often does not mean losing sight of the painting’s content altogether (not every mark is experienced as meaningless when some of them are), it can be easy to overlook the fact that we have transitioned from one sort of experience to another. We might then cast these experiences as the two components of seeing-in, and in so doing conclude that seeing-in involves an appreciation of the picture’s surface that is separate from our awareness of its content. That would be a mistake. The experience of seeing O in P with which I am concerned is the experience that in this example is transitioned to: the experience of seeing P’s (whole) content in it. It is this experience that I contend, with Wollheim, is presented to us as a single experience.⁴

In this respect, seeing O in P resembles seeing O in the flesh, or seeing P as meaningless, and differs from a multi-modal experience such as being at the ballet. Whereas seeing O in P, seeing O in the flesh and seeing P as meaningless are all experienced as single experiences, the experience of taking in a ballet is not. When we visit the ballet, we are typically aware of multiple things going on at once, and we are aware of them in different ways. For instance, we see the dancers on the stage, and we hear the music from the orchestra pit. To be sure, we might be aware of these experiences as affecting each other in various interesting ways. For example, hearing the music might affect whether we perceive the dancers as joyful, or tragic, or whatever, and seeing the dancers might make different parts of the music salient, or affect whether it is heard as staccato or not. Nonetheless, for all their possible interdependencies, these experiences are presented to us as distinct if mutually inflected visual and auditory experiences. Seeing O in P is not like this. It does not present to us as resolving into different component experiences, even interdependent ones, any more than seeing O in the flesh or seeing P as meaningless do.

The fact that seeing-in presents itself as a single experience can seem hard to square with the fact that it is twofold. How can seeing-in involve two distinct modes of visual awareness if we experience it as a single experience? Ordinarily, when we undergo two distinct modes of visual awareness, for instance when we visualize an absent object at the same time as perceiving the room around us, the result is a manifestly composite experience with separate experiential components. What is it about seeing-in that makes it different? The answer to this question – and this is the paper’s main claim – is that unlike most modes of visual awareness, the modes of visual awareness operative in seeing-in, p-vision and o-vision, are connected in a distinctive way. Specifically, they are what I’ll call converse modes of visual awareness:

converse modes of visual awareness: m₁ and m₂ are converse modes of visual awareness if, and only if, (a) m₁ and m₂ are distinct modes of visual awareness, and (b) being experientially aware of m₁-seeing something x is the same as being experientially aware of m₂-seeing something else y.

The second clause (b) states something very strong. It states something much stronger than that m₁-seeing something x is necessarily accompanied by, or entails, m₂-seeing something else y, and vice versa. It states that being experientially aware of m₁-seeing something x is the very same thing as being experientially aware of m₂-seeing something else y. In other words, there is no difference from the subject’s perspective between m₁-seeing something x and m₂-seeing something else y. From that perspective, they present as one and the same. This puts us in a position to state more precisely now what is meant by saying of an experience E, in which m₁ and m₂ are the only operative modes of visual awareness, that it presents to the subject as a single experience: in the case that m₁ and m₂ are distinct, it comes to the claim that they are converse modes of visual awareness.⁵

It should be emphasized that although (b) states something very strong, it does not follow from the fact that m₁ and m₂ are converse to one another that they are the same mode of visual awareness: (b) is consistent with (a). If (b) were the claim that being experientially aware of m₁-seeing something x is the same as being experientially aware of m₂-seeing that very thing x – call that claim (b′) – it would follow that m₁ and m₂ were the same mode of visual awareness. However, that is not the claim that (b) makes. It states that being experientially aware of m₁-seeing something x is the same as being experientially aware of m₂-seeing something else y. This is not the same as (b′), and does not entail it either.

If this paper has a central contention, it is that p-vision and o-vision are converse modes of visual awareness. In a way, that is a slightly misleading statement, because in holding that p-vision and o-vision are converse modes of visual awareness, I do not take myself to say more or less than that seeing-in is twofold and experienced as a single experience. The claim that p-vision and o-vision are converse modes of visual awareness is just another way of putting Wollheim’s twin observations about seeing-in. However, it is a way of putting those observations that helps to quell any puzzlement about their combination that we might initially have. In other words, the aim of the foregoing remarks has been to replace the question ‘How can seeing-in be both twofold and experienced as a single experience?’ with the statement ‘Given that seeing-in is both twofold and experienced as a single experience, p-vision and o-vision must be converse modes of visual awareness’.

This is a helpful move insofar as understanding is better than puzzlement. But it also helps to make manifest various intriguing facts about the nature of pictorial experience. Apart from its consequences for analyses of seeing-in (§5), perhaps the most significant of these is the fact that p-vision cannot be what I’ll call ‘run-of-the-mill perception’ – that is, the mode of perception that we have of tables and chairs and the like. It cannot be that mode of perception because being experientially aware of perceiving something in the way that tables and chairs are perceived is not the same as being experientially aware of undergoing a non-perceptual mode of visual awareness of something else. That identity is obviously false because perceiving tables and chairs does not necessarily entail being non-perceptually aware of anything. As a result, one of the more interesting consequences of the fact that p-vision and o-vision are converses is that perception has to be thought of as a species of visual awareness that divides into more determinate kinds (Figure. 1). Our perception of ordinary, non-depictive objects involves one species of perception, namely run-of-the-mill perception, and our perception of surfaces whose contents we grasp involves another.⁶

This gives us a way of characterizing the transition from seeing a picture, P, as meaningless, to seeing something, O, in it. When P is seen as meaningless, we perceive it in the run-of-the-mill perceptual mode. Coming to see what P depicts involves shifting from that perceptual mode to a different species of perception, namely p-vision. This explains the change in phenomenology that making the transition involves. Everybody agrees that the transition from seeing P as meaningless to seeing its content in it involves a shift in the phenomenology of our experience of P. For instance, Dominic Lopes states that ‘the design of Picasso’s Portrait of Daniel-Henry Kahnweiler acquires a new appearance once you see Kahnweiler in it. Otherwise disparate regions of the picture’s surface come to look organized into a whole that looks responsible for depicting Kahnweiler’ (Lopes 2005, 40). Similarly, Robert Hopkins has said that ‘when we see something, O, in a surface P, we experience P differently from when we see nothing therein’ (Hopkins 2003, 156; see also Hopkins 1998, 15). The distinctive point here is that this shift in our phenomenology is due to a shift in the mode of perception that our awareness of the surface takes. Our experience of P before the transition differs phenomenologically from our experience of P after the transition because the mode of our perceptual awareness of P has changed.

It might help to apply these rather general claims to a particular case. Suppose that you are looking at a reproduction, P, of Jastrow’s duck-rabbit picture. (This bold-faced ‘P’ is not schematic; it refers to a particular, concrete picture that we are imagining is before you.) At t-1, you experience P as meaningless, and at t, you see a rabbit in it. This transition amounts to switching from one perceptual mode of awareness of P to another. Whereas you run-of-the-mill perceive P at t-1, you p-see it at t. Moreover, this switch is all that the transition comes to. True, you o-see a rabbit at t that you did not 0-see at t-1, but your experience of o-seeing a rabbit at t is no other than your experience of p-seeing P at t. Per our definition of a single experience, these experiences are a single (i.e., the same) experience because they present to you as being one. Your experiential awareness of p-seeing P at t is the same as your experiential awareness of o-seeing a rabbit at t. You are experientially aware of p-seeing P and in that very fact experientially aware of o-seeing a rabbit. In other words, your experiences of p-seeing P at t and o-seeing a rabbit at t present to you as the same experience. Thus, while it is true that you o-see a rabbit at t that you did not 0-see at t-1, this change is not additional to, but the same as, the transformation of your perception of P from a run-of-the-mill perception at t-1 to a case of p-vision at t.

Suppose now that at t+1 you see a duck in P instead of a rabbit. Unlike the first transition from seeing P as meaningless to seeing a rabbit in it, this transition doesn’t involve undergoing a shift in any visual mode; you continue to p-see P, and you continue to o-see something. Rather, it is the content of your experience that has now changed. Whereas you previously saw a rabbit in P, you now see a duck in it. A similar shift in content, but not in mode, takes place when you turn, at t+2, to a numerically distinct photocopy of P – let’s call it ‘P*’ – and see a duck in it, instead of seeing one in P. Seeing a duck in P* is a distinct experience from seeing a duck in P not because the operative modes of awareness in them are different, but because the contents of these experiences are not the same. In one case, a duck is seen in P, and in another case, a duck is seen in P*.⁷

It is easy to lose our heads here. It is natural to think that the descriptions ‘your experience of P at t’ and ‘your experience of a rabbit at t’ refer to two different experiences, each of which is a component part of your (whole) experience of seeing a rabbit in P at t. In his later work, Wollheim stressed that seeing-in does not resolve into any such components, but that it is a single experience with two distinguishable but inseparable aspects, or folds (Wollheim 1987). There is the ‘configurational fold’, in which the marked surface is given, and the ‘recognitional fold’, in which the absent object or scene is seen. Although Wollheim insisted that seeing-in is not the simultaneous occurrence of two distinct experiences, some people have, I think, made the mistake of treating these two ‘folds’ as if they were the proper experiential parts of seeing-in (Hopkins 2003, 2010; Budd 2008). In fact, I am not sure that this mistake is entirely avoided by Wollheim himself. Wollheim tells us that each ‘fold’ is analogous to a certain experience: the ‘configurational fold’ is analogous to seeing the surface without seeing anything in it, and the ‘recognitional fold’ is analogous to seeing the absent object or scene face-to-face. In analogizing each ‘fold’ to a certain experience, these statements encourage the idea that the two ‘folds’ of seeing-in are at the very least experience-like. It is then hard to avoid the thought that Wollheim ducks a legitimate question when he refuses to say in what respect each ‘fold’ is analogous to the cited experiences. Wollheim asserts that ‘we get not so much into error as into confusion if…we ask how experientially like or unlike each aspect is to the analogous experience’ (ibid., 46), but it is far from clear why that is so if the analogies he offers are accepted.

In my view, this series of remarks represents a misstep on Wollheim’s part in an otherwise insightful passage. I am happy to say that your experience of seeing a rabbit in P at t has two aspects, or folds, but all this amounts to is the observation that the experience is twofold: you are visually aware of P, and you visually discern a rabbit in it. Because the experience is twofold, a complete specification of it requires an expression such as (a)

• (a)

  Your experience of seeing a rabbit in P at t

in which both P and a rabbit are mentioned. It is thoroughly misleading to suggest that in contrast to (a), (b) and (c)

• (b)

  Your visual experience of a rabbit at t

• (c)

  Your visual experience of P at t

denote one or other aspect of the experience that (a) denotes. The right thing to say is not that (b) and (c) each denote a different aspect of your experience, but that they each denote your experience in ways that bring out different aspects of it – that is, they each denote your entire experience of seeing a rabbit in P at t. They ‘bring out’ different aspects of this experience in the sense of specifying different parts of its content while falling short of a complete specification of it: (b) does not mention what you see a rabbit in, and (c) does not mention what you see in P. On this view, there is only one experience to which the experience of seeing P as meaningless, and the experience of seeing a rabbit face-to-face, can be analogized. This is the (whole) experience of seeing a rabbit in P. Both experiences can be analogized to this experience, but the analogy in each case is rather loose. Seeing P as meaningless is analogous to seeing a rabbit in P insofar as both involve being perceptually aware of P, and seeing a rabbit face-to-face is analogous to it insofar as both involve being visually aware of a rabbit.

Let me sum up the results of this section. Seeing-in is a twofold experience. It involves being aware of two different things in two different visual modes at the same time. By stipulation, these modes of visual awareness are p-vision, which is perceptual, and o-vision, which is not. Seeing-in is also a single experience, which is to say that it is experienced as a single experience. It is another way of putting these observations to say that p-vision and o-vision are converse modes of visual awareness: being experientially aware of p-seeing something x is the same as being experientially aware of o-seeing something else y. As a result of this fact, p-vision cannot be run-of-the-mill perception. The transition from seeing P as meaningless to seeing something (anything) in it involves undergoing a shift in one’s mode of perceptual awareness of P from run-of-the-mill perception to a different species of perceptual awareness. This shift in one’s mode of perceptual awareness of P exhausts the transition from seeing P as meaningless to seeing something in it, because although the latter state involves a new non-perceptual mode of visual awareness, namely o-vision, being aware of o-seeing something is the same as being aware of p-seeing something else. Thus, in the case that S sees a rabbit in P at t, there is no difference between S’s experience of o-seeing a rabbit at t and S’s experience of p-seeing P at t. Both experiences are S’s experience of seeing a rabbit in P at t, which the sentences ‘S perceives P at t’ and ‘S is visually aware of a rabbit at t’ allude to in different less-than-specific ways.⁸

3. A Potential Objection

In this section, I want to consider a prima facie objection to the preceding account. The account maintains that to be experientially aware of p-seeing something x is the same as being experientially aware of o-seeing something else y. Per our definition of a single experience, it follows from this contention that if S sees O in P, S’s experience of P is the same as S’s experience of O. Both experiences are S’s experience of seeing O in P. The challenge to the proposal is this. We naturally say things like ‘S sees O by seeing P’. By contrast, the converse thought is awkward: ‘S sees P by seeing O’ does not seem right. Doesn’t this asymmetry suggest that S’s experience of P cannot be the very same as their experience of O? After all, if S’s experience of P were the same as their experience of O, we should be able to intersubstitute the expressions ‘S sees P’ and ‘S sees O’ in every extensional context in which they occur. As the example demonstrates, we apparently cannot.

In responding to this objection, let me start by saying that I accept the asymmetry of the word ‘by’ in this context. It does not follow from the truth of ‘S φs by ψ-ing’ that ‘S ψs by φ-ing’ is also true. I also accept that ‘S sees O by seeing P’ is felicitous and that ‘S sees P by seeing O’ is not. What I do not accept is the suggestion that these observations jointly entail that S’s experience of P is different from S’s experience of O. This would-be entailment rests on an optional assumption, which is that the ‘by’ statements in question express a relation between experiences. This reading of ‘by’, while natural, is not compulsory. Indeed, notice that the phrase ‘S sees P’ does not even occur in the sentence ‘S sees O by seeing P’; likewise, the phrase ‘S sees O’ does not occur in the sentence ‘S sees P by seeing O’. This fact should caution against the idea that the ‘by’ statements express a relation (asymmetric or otherwise) between S’s experiences of O and P, respectively.⁹

If ‘by’ is not a relational term between experiences, what do these sentences express? Here is what I suggest. The sentence ‘S sees O by seeing P’ attributes a causal role to P. It communicates that S has a visual impression of O (specifically, she sees O in P) as a result of entering into a particular sensory relation with P. A certain amount of light falls onto P and is reflected onto S’s retinas, and S sees O (in P) as a result. This claim is true, and the sentence is accordingly felicitous. (It is important to stress that the causal role is attributed to P itself, and not to S’s experience of P. S’s experience of P is S’s experience of O, and so cannot play any causal role in its production.) The sentence ‘S sees P by seeing O’ invokes the same causal role but this time attributes it to O. It communicates that S sees P because a certain amount of light is reflected off O and onto their retinas. This claim is false, and the sentence is correspondingly infelicitous. O might play some causal role in the fact that S sees P – it might be a partial cause of P’s existence – but it does not play the causal role that the sentence-form ‘S sees…by seeing…’ discusses. This way of understanding that sentence-form accommodates the asymmetry between the sentences ‘S sees O by seeing P’ and ‘S sees P by seeing O’ in a way that is consistent with the fact that S’s experience of P is the same as S’s experience of O. That single experience occurs because S enters into a particular sensory relationship with P and not O.

4. ‘Nested’ Seeing-in

Before situating this paper’s claims with respect to the existing literature, I want to briefly consider the very interesting case of seeing pictures of pictures (Lopes 2005; Hopkins 2010; Terrone 2013; Kulvicki 2006, 2020). Such cases give rise to what are sometimes referred to as ‘nested’ cases of seeing-in: we see x in y, which is itself seen in z. Take Vermeer’s Woman Holding a Balance (Figure 2). In this painting, a picture of the Last Judgement is seen hanging on the wall behind a woman with a set of weighing scales. One of the virtues of thinking about seeing-in through the lens of its converse modes of awareness is that it helps us to get straight on the nature of such experiences. A brief aside will help to demonstrate this.

Figure 2: Johannes Vermeer, Woman Holding a Balance (1662-3). National Gallery of Art, Washington, D.C.

Given this paper’s claims, we might ask ourselves: Does the mode of our awareness of the Last Judgement, which is seen in a picture behind the woman, differ from the mode of our awareness of the woman? We can say for certain that both the Last Judgement and the woman are o-seen: both are seen in arrays of ink-marks. But does o-vision divide into two different subspecies according to whether the object o-seen is seen (like the Last Judgement) in a depicted picture, or (like the woman) in a real one? This question can be asked another way. Let’s call the collection of ink-marks that correspond to the depicted frame ‘F’. Suppose we go from seeing F as depicting a picture frame to seeing it as depicting a window frame, so that the Last Judgement is first seen in a picture in the painting, and subsequently as belonging to the woman’s world. Does this transition involve a shift in the mode of our awareness of the Last Judgement from one species of o-vision to another?

The answer to this question is, I think, ‘No’. The reason for thinking that the transition from seeing O in the flesh to seeing it in a picture results in a shift in the mode of our visual awareness of O is that making the transition introduces an additional mode of visual awareness (namely, p-vision) without thereby multiplying the number of experiences that we are aware of having (cf. fn. ⁷). This pressure does not apply here. Neither the transition from simply seeing the Last Judgement in the painting to seeing it in a picture in the painting, nor the opposite transition, introduces a secondary mode of visual awareness that was not operative before. In neither case are we newly aware of anything in an additional visual mode. Thus, whether F is seen as depicting a picture frame or as depicting a window frame, the mode of our visual awareness of the Last Judgement is no different from the mode of our visual awareness of the woman.

This has an interesting consequence, which is that it is somewhat misleading to describe seeing a picture in a picture as a ‘nested’ case of seeing-in. That way of describing it implies that the experience involves one case of seeing-in somehow embedding another structurally identical case of seeing-in: ‘we see x in y, which is itself seen in z’. It is not clear what such ‘embedding’ talk is supposed to come to, but in any case, the sense in which x is seen in y is quite different from the sense in which y is seen in z. This comes out in the fact that the transition from seeing the Last Judgement in a depicted picture to seeing it through a depicted window is not at all like the transition from seeing the Last Judgement in a real picture to seeing it through a real window (i.e., face-to-face). While the latter transition involves a shift in the mode of one’s visual awareness of O (from o-vision to run-of-the-mill perception), the former does not. In the example given, the former transition is the result not of switching visual modes, but of switching the concept (from picture frame to window frame) under which the depicted object corresponding to F is brought. It does not involve ceasing to be aware of anything. This distinguishes it from the transition from seeing the Last Judgement in a real picture to seeing it through a real window, which involves ceasing to be aware of some ink-marks. When we go from seeing the Last Judgement in a depicted picture to seeing it through a depicted window, we are aware of the same things throughout, but one of those things (the object that F depicts) is brought under a new concept.

5. Comparisons to the Existing Literature

In this last section, I want to briefly situate this paper’s claims with respect to some of the existing literature. Lots of analyses of seeing-in have been defended over the years. The three most prominent accounts are transparency theories, experienced resemblance theories, and Kendall Walton’s ‘make-believe’ theory. According to transparency theorists (Kulvicki 2009; Newall 2015), to see O in P is to undergo a perception of O as lying behind P; it appears to you as though you see through the picture’s surface to a scene behind it. According to experienced resemblance views (Peacocke 1987; Budd 1993; Hopkins 1995, 1998), seeing O in P is an experience of likeness; P is experienced as resembling O in a certain respect, such as ‘outline shape’. According to Walton’s ‘make-believe’ theory (Walton 1990), seeing O in P involves imagining of your perception of P that it is actually a perception of O. Despite their differences, these views are united by the shared task of saying something substantive about what seeing-in is. They try to characterize the relation in which O and P are experienced as standing in terms that are intelligible independently of the concept of seeing-in. Transparency theorists posit a spatial relation, proponents of experienced resemblance posit a resemblance relation, and Walton posits a rather more complicated relation according to which O is imagined as the content of a perception that is (and is experienced as being) actually of P.

My purposes in this paper have been different. I have not tried to give an analysis of seeing-in in independently intelligible terms. In particular, I have not given a reductive characterization of what p-vision and o-vision are. All I have said is that p-vision, by definition, is the perceptual mode of visual awareness that we have of a marked surface (any marked surface) when we see an object or scene (any object or scene) in it, and o-vision, by definition, is the non-perceptual mode of visual awareness that we have of an object or scene (any object or scene) when we see it in a marked surface (any marked surface). Clearly, these definitions make essential use of the concept of seeing-in. Rather than give an analysis of them, I have demonstrated what must be true of p-vision and o-vision given that seeing-in is a twofold but single experience. It is easy to think that these features of the experience conflict but actually all they do is tell us something interesting about p-vision and o-vision, which is that they are converse modes of visual awareness: being experientially aware of p-seeing something x is the same as being experientially aware of o-seeing something else y.

Although this is not itself an analysis of seeing-in, it does significantly alter the landscape of possible analyses of seeing-in. For instance, I began with the assumption that o-vision is a non-perceptual mode of visual awareness. (This was explicit in my definition of twofoldness.) This rules out transparency theories, which maintain otherwise. Moreover, in holding that seeing O in P means undergoing a distinctive mode of perception of P and that is all, I cannot obviously accept Walton’s imagination theory either, according to which seeing-in involves a higher-order imaginative act that works on a first-order perception of P. For this view to be consistent with the claims defended here, being experientially aware of p-seeing something x would have to be the same as being experientially aware of imagining of that perception that it is of something, y, other than x. It is not clear (to me, anyway) that this is coherent. Because the imaginative act works on the perception in a way that, as it were, transforms it, I cannot see how the act of imagination can then be experienced as being the very same experience as the (pre-transformation) perceptual act. Hopkins is thus wrong to suggest that ‘Walton’s view does not decompose [seeing-in] into a component in which we experience the marks, and another in which we are presented with the depicted scene’ (Hopkins 2021). Indeed, Walton’s own pronouncements that our experience of the marks ‘integrates’, or ‘amalgamates’, with our experience of the depicted scene suggests that he regards these experiences as distinct (if inseparable) components of seeing-in.¹⁰

Interestingly, and in contrast to the preceding analyses, experienced resemblance views do not obviously conflict with this paper’s conclusions. On the contrary, proponents of these views can be thought of as accepting those conclusions while at the same time supplementing them with a reductive account of p-vision and o-vision. Let me spell this out. Proponents of experienced resemblance views hold that seeing O in P is a bit like perceiving someone to resemble their late mother: it involves perceiving P to resemble O (which is absent) in a certain respect. (Hopkins suggests that this respect is ‘outline shape’, but we needn’t worry about the details here.) Although proponents of this view do not typically present themselves as saying something about the modes of visual awareness operative in seeing-in, interpreting them this way yields an attractive construal of their proposal. That is, we can think of them as saying something independently intelligible about what o-vision and p-vision are: p-vision is the mode of perception that we have of an object when it is perceived as resembling something, and o-vision is the mode of non-perceptual visual awareness that we have of an object when something is perceived as resembling it.¹¹ Calling these modes of awareness, respectively, ‘ER_ing-vision’ and ‘ER_ed-vision’ (‘ing’ for ‘resembling’, and ‘ed’ for ‘resembled’), the experienced resemblance view can be put like this: p-vision is ER_ing-vision and o-vision is ER_ed-vision, or seeing O in P is ER_ing-seeing P and ER_ed-seeing O. Interpreted thus, this view accords with the paper’s conclusions if, and only if, ER_ing-vision and ER_ed-vision are converse modes of visual awareness. And very plausibly, they are: being experientially aware of ER_ed-seeing something x is plausibly the same as being experientially aware of ER_ing-seeing something else y.

This is not the place to offer a proper assessment of experienced resemblance views. So, let me end by saying that for my own part, I am not sure that the endeavour to say what p-vision and o-vision are in independently intelligible terms is a good one.¹² I am inclined to think that the fundamental way of characterizing the relation in which O and P are experienced as standing when O is seen in P is to say that we see O in P. This might sound defeatist, but one attractive consequence of holding this is that o-vision turns out to be unique to the pictorial context broadly construed, i.e., to cloudy skies, petrol spills, pieces of burnt toast and the like, as well as to intentionally constructed configurations of ink-marks. Whereas the experienced resemblance theorist maintains that John’s mother is experienced in the same mode of visual awareness whether we see John as resembling her or see her in a portrait, I prefer the idea that her portrait gives us a distinctive kind of awareness that cannot be had by other means. We can perceive his mother in the flesh, we can visualize her, we can perceptually recall her, we can see objects as resembling her, and (in addition) we can see her in pictures and oil spills.¹³