This paper outlines what I will call a ‘perceptual account’ of depiction. A perceptual account differs from other accounts of depiction in that it takes at face value the claim that we see things in pictures. Like an experiential account of depiction,Footnote 1 a perceptual account involves that experiencing a picture as a picture of something—or seeing something in a picture, as one often puts it colloquially—is to have a visual experience of something other than the picture. But the claim is not that one thereby visually experiences something absent, as defenders of an experiential account hold. Rather, a defender of a perceptual account claims that one visually perceives something present in the realm of the picture when experiencing a picture as a depiction. While this claim may come across as rather far out at first, I think it can be made good sense of. In fact, I think several versions of the perceptual account are available, each with different suggestions as to what type of entity can be present in the realm of a picture.

The first section motivates perceptual accounts of depiction generally and distinguishes this type of account from some central accounts of depiction. Later sections, in turn, are concerned with distinguishing different versions of the perceptual account. Section 2 examines Wiesing’s (2010) perceptual account with respect to the constraints it puts on the nature of the thing, T, that is seen and present in a picture. These constraints are challenged in Sects. 3, 4 and 5, where three different suggestions as to what Ts may be are considered, namely the suggestions that Ts may be material objects (Sect. 3), visual appearances (Sect. 4) or universals (Sect. 5). Section 6 concludes.

1 A motivating puzzle

Let me start by making two fairly natural observations about pictures, presence and visibility. Observe first that when one is in Paris one can see the Houses of Parliament in a picture. For instance, one may visit Musée d’Orsay and see the Houses of Parliament in one of Monet’s paintings.Footnote 2 But observe secondly that if one is to see the Houses of Parliament, and not merely see it in a picture, then one needs to go to London. Indeed, this is one of the typical reasons why people go to London; they wish to see the Houses of Parliament. The relevant difference between Paris and London in this regard, it is natural to suggest, is that the Houses of Parliament is located in London, but not in Paris. The Houses of Parliament is present to one when one is oneself in London. These two observations lead to a contradiction when put together. In general:

  1. (1)

    When O is seen in a picture, O need not be present.

  2. (2)

    In order to see O, O must be present.

  3. (C)

    So, in order to see O in a picture, (i) O must be present and (ii) O need not be present.

How can we avoid the contradiction in (C) between (i) and (ii)? Answering this question provides a useful way of distinguishing the perceptual account from other accounts, since different accounts of depiction yield different answers to the question. I will mention a few alternative ways in which the puzzle can be dissolved, before we look at how a perceptual account responds to it. First of all, however, let me put to rest some potential worries about the puzzle as such.

While I take it that (1) is uncontroversial, it is in good order to explain why also (2) is fairly natural to accept. Note that some theorists let the plausibility of (1) serve to rule out (2). This seems to be involved in Walton’s (1984) photographic realism, according to which photographs (although not pictures generally) are a type of visual aid that enable us to, e.g., ‘see, quite literally, our dead relatives themselves when we look at photographs of them’ (ibid., p. 252). On this view, our relatives need not be present in order to be seen. However, photographic realism has few defenders, and in this sense it is fairly natural to accept (2).

Note, moreover, that (2) may seem more plausible when observing that it is only a necessary and not a sufficient condition for seeing. Also when one has an illusory perceptual experience the object experienced is present to one. For instance, one may be looking at the Houses of Parliament when it is illuminated by coloured light such that the building, rather than appearing to be viewed under special lighting conditions, appears to have a different colour than what it really has. In this case, the Houses of Parliament is present to one, but it is not seen, given that seeing or visually perceiving involves veridicality. Rather, it is visually experienced in a non-veridical way.

Finally, note that the notion of presence employed in stating both (1) and (2) is flexible and primitive. What counts as present depends on the circumstances. For instance, it need not be immediately before one. A mountain top several hundred metres away from the perceiver, or a distant star long since deceased, may count as present—even if a mouse in the same location may not so count. While it is difficult to conceive of a general rule that adjudicates upon what counts as present in every case, we seem to have a clear enough grasp of what counts as present without such a rule. Hence we can take the notion as primitive.

Now, given that (1) and (2) are fairly natural observations, the puzzle presented above is in need of a solution. As I mentioned, we can distinguish between different accounts of depiction by looking at their response to the puzzle. For instance, on a resemblance account of depiction, according to which a picture depicts O by resembling it,Footnote 3 the contradiction in (C) is easily avoided. For according to this account, we do not see O itself, but only something that resembles O, when we see O in a picture. Similarly, on a semiotic account like Goodman’s (1976) according to which pictures depict by means of symbolic convention, the contradiction can be avoided by explaining that we, in a sense, “read” pictures. As when one is reading, there is no reason to suppose that what is read about, or what the picture is “about”, namely O, is present. These two responses both involve the idea that seeing O in a picture is to see something that is connected to O (either by resemblance or by symbolic convention) but not O itself. This makes the puzzle dissolve, since it means that (1) and (2) have different subject matters because they fail to concern the same object. In short, it is denied that seeing O in a picture involves seeing O.

While it may be right to deny that seeing O in a picture involves seeing O, it seems we should nevertheless maintain that seeing O in a picture involves having a type of visual experience of something other than the physical object that is the picture. For it is a central trademark of representational pictures that they facilitate visually experiencing something other than themselves. Resemblance accounts and semiotic accounts fail to accommodate this point; they only make room for the idea that we have a visual experience of the physical object that is the picture, or the marks on the canvas. By contrast, so-called experiential accounts of depiction emphasise that seeing O in a picture is to have a type of visual experience of something absent, namely the depicted object, O. Following standard practice, we may use the notion of ‘seeing-in’ to refer to this type of visual experience. Experiential accounts generally construe seeing-in as a visual experience that involves awareness of both the picture and the depicted object, where these two components of the experience cannot be had alone without changing the phenomenology of the components.Footnote 4 This facilitates responding to the puzzle by replacing (1) with the following:

(1′) When O is seen-in a picture, O need not be present.

By replacing (1) with (1′), experiential accounts are, like resemblance accounts and semiotic accounts, construing the two observations in (1) and (2) as having different subject matters. Rather than concerning different objects, experiential accounts construe (1) and (2) as concerning different types of experience; (2) concerns seeing face to face, whereas (1) concerns seeing-in. Hence, there is no conflict between the observations, and the puzzle dissolves.

A central and challenging task for experiential accounts is to explain the nature of seeing-in and how it differs from seeing. This challenging task is avoided if we instead adopt a perceptual account of depiction, which can now be introduced as a third way of dissolving the puzzle. A perceptual account takes at face value the claim that we see things in pictures. There is no need for postulating a special type of visual experience involved in seeing pictures. Instead, a perceptual account says that what is seen in a picture is present in it. This means that the requirement in (2) is accepted as applying also to what we see in pictures. It is immediately clear that unless this application is to lead to the contradiction in (C), one must allow that what is seen in a picture can differ from what the picture is a picture of, i.e. the depicted object O. Let me use the letter ‘T’ as a placeholder for whatever thing is seen and present in a picture. A perceptual account’s strategy for avoiding the contradiction in (C) can then be explained as replacing premise (1) with the following:

(1″) When T is seen in a picture, O need not be present.

The idea is that the depicted object, O, is not identical to what is seen and present in a picture; what is seen is instead T. So, even though it is maintained that whatever is seen must be present when it is seen, as premise (2) states generally for any object of sight, O, combining this requirement with premise (1″) does not result in imposing contradictory requirements on the presence of what is seen in pictures, as in (C). Instead we arrive at the following conclusion:

(C″) In order to see T in a picture of O, (i) T must be present and (ii) O need not be present.

Let me now summarise how a perceptual account’s response to the puzzle differs from that of resemblance accounts, semiotic accounts and experiential accounts. This will help to identify the main explanatory tasks facing a defender of the perceptual account.

Like resemblance accounts and semiotic accounts, perceptual accounts involve the idea that one sees something different from but connected to O when one sees O in a picture. But what is thus connected to O is not, as resemblance accounts and semiotic accounts suggest, the physical object that is the picture, or the marks on the canvas. Rather it is something present in the picture, i.e. T. Since perceptual accounts postulate this connection between Os and Ts, it is clear that a central explanatory task is to describe the nature of the connection, similarly to how resemblance accounts and semiotic accounts describe the connection between Os and pictures.

Since T is not the physical object that is the picture or the marks on the canvas, perceptual accounts can, by contrast to resemblance accounts and semiotic accounts, accommodate the natural idea that seeing O in a picture involves having a visual experience of something other than the picture. In this respect, perceptual accounts are on a par with experiential accounts. But instead of claiming, as experiential accounts do, that the other thing experienced is something absent, namely the depicted object, O, perceptual accounts claim that it is something present in the realm of the picture, namely T. It has not yet been specified what kind of thing T is. One may even wonder if any entity at all can play the role I have described Ts as having. Hence, prior to accounting for the nature of the connection between Ts and Os, the most pressing task for perceptual accounts is to determine what sort of entities Ts are.

This task will be the primary focus in what follows. Different versions of the perceptual account will be considered, each with their own choice of the type of entity T is. Each of these choices restricts how we may account for the connection between Ts and Os. However, before discussing the alternative choices of Ts, it is useful to first look at a version of the perceptual account recently developed by Wiesing (2010). Wiesing does not focus on determining what sort of entity T is, but he puts certain constraints on the nature of Ts. I identify these constraints in the next section. The constraints constitute a useful starting point for the discussion in Sects. 3, 4 and 5.

2 Wiesing’s perceptual account

Let us first observe that Wiesing’s account of depiction possesses the central features of a perceptual account mentioned in the previous section.Footnote 5 Firstly, Wiesing acknowledges the threefold distinction, set out above, between T, O, and the picture. Adopting Husserl’s (1913) framework, Wiesing explains this threefold distinction by using the notions of image carrier (Bildträger), image subject (Bildsujet) and image object (Bildobject). Secondly, Wiesing’s account involves that the third element, the ‘image object’, or ‘T’ as I will continue to refer to it, is present and seen in the picture. More specifically, Wiesing holds that Ts are ‘artificially present’ in pictures, by which he means that they are merely visibly present, and not present in ways detectable by senses other than sight. In this way, Wiesing’s account respects the idea that seeing Ts in pictures is subject to the requirement in (2), i.e. that whatever is seen must be present. Thus, Wiesing’s account has the general features of a perceptual account. Let me us now proceed to identify four constraints Wiesing puts on the nature of Ts. Only the first of them is mandatory for a perceptual account.

The first constraint is a negative one, arising from comparison with semiotic accounts of depiction. Wiesing emphasises that acknowledging three elements in one’s account of depiction is not special to perceptual accounts, for also semiotic accounts postulate three elements. But there is a difference with respect to how the three elements are conceived of on these two types of account. On a perceptual account, the first element, i.e. the picture, is simply understood as what hangs on the wall, gathers dust, can be shed light on, sold, bought, and so on. Semiotic accounts, by contrast, add that the picture is also a sign; it has the function of standing for something other than itself. Concomitantly, the depicted object, O, is on semiotic accounts always a referent, and the third element in representation is a content in virtue of which the picture refers to O. This commitment to signs, referents and contents is avoided by the perceptual account. Although the perceptual account can make room for the fact that that pictures sometimes function as signs, this is not regarded as part of what it is for a picture to depict. Thus, Wiesing emphasises that ‘what distinguishes the approach based on perception [from the semiotic account] is the idea that we can give up on the idea that images necessarily refer’ (Wiesing 2010, p. 22–23). Wiesing’s first negative constraint on Ts is thus that they are never necessarily contents, in virtue of which the picture refers. This negative constraint is a mandatory constraint on all versions of the perceptual account, since violation of it amounts to endorsement of a semiotic account.

Let us now turn to those of Wiesing’s constraints on Ts that are special to his version of the perceptual account. As briefly mentioned already, Wiesing holds that Ts are ‘artificially present’ in pictures. They are merely visibly present, and not present to other senses in addition. As Wiesing puts it, T is ‘not completely present’: ‘For it is indeed exclusively visibly present and not present in that complete way accessible to all senses with which we are familiar from real things’ (ibid., p. 50). Thus, by contrast to experiential accounts, Wiesing holds that what is seen in a picture is not something absent, but rather something which is present in this special manner.

Wiesing understands the constraint concerning artificial presence in a particular way. He thinks the constraint involves that ‘things in images are exclusively visible and never collect dust’, and that ‘what we see in the picture has no material substance’ (ibid., p. 20). So, instead of thinking of the constraint as being, merely, that there is something special about the way in which Ts are present, i.e. that they are only present to the eyes, Wiesing seems to regard the constraint as involving that the nature of Ts is special, i.e. that Ts are ‘exclusively visible’ and hence not material objects. It is this aspect of their nature which is responsible for the fact that they are only artificially present, he seems to think. As far as I can understand, however, the claim that Ts are exclusively visible and hence not material objects is an additional constraint to that concerning artificial presence. For it does not seem to be ruled out in advance that something which is not a pure visible, such as a material object, can be purely visibly present to one. By contrast to Wiesing, therefore, I count the constraint concerning artificial presence and the constraint concerning pure visibility as two independent constraints.

Also the fourth of Wiesing’s constraints on Ts is in his view interconnected with the two constraints just mentioned. Following Husserl, Wiesing claims that Ts are intentional objects. They are objects that become visible in a picture when someone sees it as a depiction, and hence they are always objects for someone, existing only as long as someone looks at the picture, he thinks. This constraint leads to that concerning artificial presence, if it is combined with the idea that we see Ts. To hold that Ts are seen and that they are always objects for someone amounts to one way of holding that Ts are merely visibly present; they cease to exist when they are no longer visibly present, and hence they are pure visibilia. Despite this connection between the two constraints, however, the constraint that Ts are intentional objects is clearly independent of the constraint that Ts are artificially present. For there can be other visible intentional objects besides those that are artificially present, and moreover, as I will return to in Sect. 4, pure visibilia need not be intentional objects.

As I have emphasised, the three constraints special to Wiesing’s account are independent of one another. They are also dispensable for a defender of the perceptual account of depiction. What versions of the perceptual account result if we dispose of them? In the next section, I look closer at the constraint that Ts are not material objects. One might think that this obviously is a constraint one should uphold, as it seems absurd to claim that material objects, like the Houses of Parliament, are present in pictures. But we shall see that, on closer inspection, the seemingly obvious reason for rejecting the claim staggers.

3 Ts as material objects

Unless we wish to endorse the contradiction in (C), it is clear that if Ts are material objects, they cannot be identical with the Os depicted by the picture. If T is O, then the same object needs to be both present (by (2)) and not present (by (1)) in the picture. However, perhaps Ts can be other actual material objects than the depicted Os? This is obviously a bad proposal, one might object, for the following reason. Suppose (for reductio) that Ts are actual material objects. It then follows that actual material objects can be located at once both in a picture and at a location in the actual world. In other words, it follows that actual material objects can be in two places at once. This is obviously metaphysically unacceptable.

On deeper reflection, however, it is far from obvious that being located in a picture and being located in London is metaphysically unacceptable. Granted, being in two places at once is metaphysically unacceptable for locations in the actual physical world (barring certain scenarios arising in quantum mechanics). But it is not at all clear that being in a picture is to be located in the actual physical world. In order to assess whether co-location in a picture and in an actual physical location is problematic, we thus need to know more about what sort of place a thing is situated in if it is present in a picture.

I envisage two general options, mirroring two uses of the notion of a picture. Borrowing Wiesing’s examples, we may distinguish between, on the one hand, claims like ‘The picture is on the wall’ or ‘The picture is torn’, and, on the other hand, claims like ‘The picture is boring’ or ‘The picture has great spatial depth’.Footnote 6 The former type of claim gets at the physical object. But it is clearly not the physical object that, for instance, has great spatial depth, since the physical object typically is flat. This distinction points to two possible interpretations of the claim that Ts are seen and present in pictures: That they are seen and present in a physical object, or that they are seen and present in the realm we become aware of when seeing a surface as a depiction. Thus, in accordance with the former interpretation, we may suggest that being present in a picture is to be located in the two-dimensional plane of the canvas, i.e. in the physical picture surface. This is obviously a location unsuited for actual material objects, since material objects are three-dimensional (or four-dimensional). So, if we think of the realm of the picture as the two-dimensional plane of the canvas, the implication that Ts are both in pictures and in actual physical locations is problematic, although not because Ts would be in two places at once.

A second suggestion, which fits with the second use of the notion of a picture, is that the realm of the picture is a three-dimensional (or four-dimensional) realm, coming into view when we see a picture as a depiction. As Alberti (1966) famously suggested in his treatise On Painting, we may think of looking at a picture as similar to looking through a window, where the plane of the canvas corresponds to the window pane. Adding to this idea, we may explain that the realm of the picture at the other side of the window pane is a three-dimensional (or four-dimensional) realm. How does this conception interact with the idea that Ts are actual material objects?

Note that on the second conception the realm of the picture is in certain ways isolated from the actual world. We spectators cannot reach into the picture, and the things in the picture cannot reach through the canvas to us. Admittedly, we may not be completely cut off from the realm of the picture. We may see the things in the picture and thus be affected by them, and we can arguably also affect the things in the picture by making changes to the canvas, for instance by drawing a moustache in the part of the canvas that depicts a person’s face. Still, at least in some respects the realm of the picture obviously remains spatially (and temporally) non-continuous with the three-dimensional (or four-dimensional) world we inhabit. This non-continuity causes problems for T’s presence in a picture, insofar as one agrees with Lewis (1986) that there cannot be disconnected spacetimes within a single world. For this would mean that the realm of the picture is not part of actuality. Since an actual material object cannot be located in a non-actual realm, this excludes Ts from being actual material objects.

However, it may be argued that due to the possibility of one-way interaction between the realm of the picture and the actual world (e.g. that spectators may be affected by things in pictures by seeing them, or that things in pictures may be changed by making changes to the picture surface), the realm of the picture is not disconnected from actuality on Lewis’s view. Alternatively, it may be argued that Lewis is wrong and that there can be island universes, i.e. worlds that have absolutely isolated spatio-temporal parts. Or, again, it might be argued that worlds can be unified in ways that are not spatio-temporal.Footnote 7 Depending on the success of such arguments, there may be a viable version of the perceptual account according to which Ts are actual material objects located both in pictures and in an actual physical location like London. What remains problematic for this view, however, is to answer the question as to which particular actual material object is seen and present in a picture of, e.g., the Houses of Parliament. We are precluded, on pain of the contradiction in (C), from opting for the most natural answer to this question, namely the Houses of Parliament itself.

Let us consider a related version of the perceptual account that avoids this problem. An alternative to letting Ts be actual material objects will be to let Ts be possible material objects. On this alternative, we can think of pictures as constituting windows to other possible worlds. Upon seeing Monet’s painting of the Houses of Parliament one is, on this view, seeing the Houses of Parliament in a different possible world. By contrast to the view that Ts are actual material objects, there is on this version of the perceptual account an obvious choice of T in each particular case: the object present and seen in the picture (T) is the depicted object (O) in a different possible world.Footnote 8 However, this version of the perceptual account will be developed in different ways depending on how one thinks of possible worlds. Each development of the view faces its own problems.

If we adopt Lewis’s (1986) view, a particularly difficult problem arises. Given his view in the background, the present version of the perceptual account says that T is O’s counterpart. So seeing, e.g., the Houses of Parliament in a picture amounts to seeing an object in a different possible world that is qualitatively identical or very similar to the actual Houses of Parliament. But: What makes it the case that seeing this merely possible object is to see a picture of the actual Houses of Parliament? Why is it not to see a picture of, e.g., a counterpart of the Houses of Parliament in a non-actual possible world, or a picture of an actual copy of it, or, for that matter, a picture of the actual Notre-Dame very much misrepresented? This is what I label the ‘connection problem’, so-called because it concerns the connection between Os and Ts. As long as we only have similarity or resemblance to appeal to, which is all we have for counterpart-relations, the connection problem cannot be put to rest.

Observe, by contrast, that if we adopt a Kripkean conception of possible worlds, the connection problem vanishes. On a Kripkean view, seeing the Houses of Parliament in a different possible world amounts to seeing it,Footnote 9 and not just something that resembles it. Thus, there is no puzzle as to how seeing T in the picture amounts to seeing a picture of O; seeing T just is to see O in a different possible world on this view. Other puzzles remain, however. For instance, it may be asked: How does the artist manage to make a window to one particular possible world, or a class of possible worlds, rather than another? What makes it the case that a spectator is seeing, e.g., the Houses of Parliament in a different possible world, and not something else, for instance the Notre-Dame? Such questions need not be unanswerable. The artist’s intention, the historical facts about the production of the work of art, the spectator’s psychology, and so on can be appealed to in answering them. But for now I leave such questions aside, in favour of exploring yet other versions of the perceptual account.

4 Ts as appearances

In the previous section, I explored lifting the constraint, put down by Wiesing, that Ts are not material objects. Recall, however, that Wiesing also puts down the positive constraint that Ts are ‘objects of pure visibility’. In light of this constraint, it is natural to suggest that Ts are visual appearances. The present section examines two ways of developing this suggestion.

One way of developing the suggestion is Wiesing’s way. He holds that Ts are intentional objects, existing always as an object for someone. But one need not hold that appearances are always objects for someone. As Hyman (2006) explains, appearances may be understood as either objective or subjective. This distinction, as Hyman helpfully frames it, concerns whether the picture is perceived to have the appearance of, e.g., the Houses of Parliament because it has that appearance, or if the picture has the appearance of the Houses of Parliament because it is perceived to have that appearance. While Wiesing clearly opts for the latter, a perceptual account can decide either way on this issue. In fact, with regard to photography in particular rather than pictures in general, there is a view developed by Martin (2012) that fits with the general characteristics of a perceptual account and that contrasts with Wiesing’s view in understanding appearances as objective.

Like Wiesing, and like perceptual accounts of depiction generally, Martin stresses that something is perceived in photographs, namely the appearance of an object. He also adheres to the distinction, central to perceptual accounts, between what is depicted (O) and what is present and seen in the picture (T). But according to Martin appearances are objective in the sense that they are observational properties of objects. This makes Martin’s view subject to a challenge that Wiesing’s view avoids. Suppose that we have before us a photograph of the Houses of Parliament. Note that also the photographed original, the Houses of Parliament, has the appearance which, according to Martin, is seen in the photograph. Thus, there arises the question: What is the difference between, one the one hand, seeing the appearance when one sees the object whose appearance it is, and, on the other hand, seeing the appearance in the picture?

Wiesing can build a difference between the two cases into his account as follows. The appearance seen in the photograph is distinguished from that of the original in that it is subjective; it is always an object for someone. Martin must explain the difference a different way. His suggestion is that experiencing a photograph is special in that it involves experiencing ‘presence in absence’. On the one hand, Martin explains, the appearance of, e.g. the Houses of Parliament, is recognised in the two-dimensional surface, and in this sense its appearance is present. But, on the other hand, the Houses of Parliament is absent; what is present is the photograph, the flat surface, not the Houses of Parliament. According to Martin, it is a virtue of his account that it makes room for this idea about ‘presence in absence’ or ‘exemplification’, for this raises the question which makes it clear that the photograph is a representation. He explains: ‘For in seeing an appearance reproduced which is, at the same time, not exemplified by what is before our eyes, there arises the question which object it is whose appearance this is?’ (ibid., p. 342).

Thus far, both Wiesing’s and Martin’s views seem to be viable versions of the perceptual account that construe Ts as appearances. The views turn out somewhat differently due to the different conceptions of appearances presupposed by each. However, a central question facing both views concerns the connection between Ts and Os. How does an appearance in a picture relate to that object, O, whose appearance is seen in it? In other words, what enables us to answer Martin’s question at the end of the previous paragraph? I think both Wiesing and Martin have trouble answering this question generally for all forms of depiction.

According to Wiesing, it is resemblance between T and O that makes seeing T in a picture amount to seeing a picture of O. The function of T, Wiesing writes, ‘consists in standing for an image subject [i.e. O], which means “that it ‘counts’ not for itself but as a ‘pictorial presentation’ of an object that resembles it”’ (Husserl in Wiesing 2010, p. 37). Note that this claim about resemblance differs from that made by defenders of a resemblance account of depiction. On a resemblance account, it is the picture, and not something present and seen in the picture (i.e. T), which is held to resemble the depicted object, O. Due to this contrast with resemblance theories, Wiesing’s claim about resemblance avoids many of the traditional problems for resemblance theories, such as the problem that pictures, which are usually flat and square marked surfaces, seem to resemble one another more than their depicta.

However, the problem with construing the relation between Os and Ts as a resemblance-relation is that it triggers what I called ‘the connection problem’. Wiesing’s view is that the appearance seen in the picture resembles O. The resemblance may admittedly be very complex; we may for instance see the appearance of a building in limestone, with a clock tower, situated by a river, having 148 windows, etc. Still, a complex appearance like this will not be sufficient for securing the connection to a particular object such as the Houses of Parliament. No matter how complex the appearance is there is no more reason for thinking that it resembles the Houses of Parliament than there is for thinking that it resembles a different building with the same visible appearance.

Martin provides a different account of the relation between Ts and Os. He provides an account particular to photographs. He envisages that, due to their mechanical means of production, photographs have an indexical link to the particular historical event where, e.g., the Houses of Parliament was photographed. In virtue of this link, he explains, the appearance present and seen in the photograph is the appearance of a particular thing. Note that the relation Martin describes thus is not one between the appearance, T, and the depicted object, O, but rather one between the appearance, T, and the physical representational object. But the connection between T and O can be explained by making a natural addition. We may add that a historical link also obtains between the representational object and O: O is the object that was photographed during the relevant historical event. This makes T and O connected via their respective historical links with the representational object.

It is doubtful, however, whether this account generalises to other forms of depiction besides photography. This depends on whether it also for non-photographic pictures generally is the indexical link to the event of their production that determines which appearance is seen in the picture. It plausibly need not be. Non-photographic pictures made by copying the appearance of a particular object can amount to depictions of other particular objects. For instance, a painter may depict the Virgin Mary by copying the appearance of a sitter. One may even think that such discrepancy between the object whose appearance is copied and the object depicted can occur in photography.Footnote 10 In any case, even if one should think such discrepancy cannot occur in photography, it at least plausibly occurs in some forms of painting. Hence it is evident that the account of the relation between Ts and Os that Martin’s view permits of does not generalise to all cases of depiction.

In conclusion, the relation between Ts and Os remains problematic on both the subjective and the objective way of developing the version of the perceptual account that construes Ts as visual appearances. Nevertheless, both views manage to distinguish between seeing the appearance in the picture and seeing it face to face; Wiesing can explain that the appearance in the picture is special because it is subjective, and Martin can explain that it is special because it involves ‘presence in absence’ or ‘exemplification’. This alerts us to a structural virtue of Wiesing’s and Martin’s accounts: They make the relation between T and O differ from the relation between T and the picture. The final, and in my view preferable, version of the perceptual account to be considered in the next section preserves this structure, while it simultaneously, and by contrast to the suggestion that Ts are appearances, makes good sense of the connection between Ts and Os. The final version construes Ts as universals instantiated by Os but not by pictures.

5 Ts as universals

The idea that pictures present us with universals or properties is well-known. For instance, Schopenhauer (1969) holds that we in art generally become aware of Platonic ideas. Also more recently it has been argued, by for instance Zeimbekis (2010), that pictures present us with properties rather than particulars. By contrast to many familiar varieties of the idea that pictures present us with properties or universals, the version of the perceptual account I envisage involves that the properties or universals with which we are presented in pictures can include, for instance, the property or universal of being the Houses of Parliament. We need to include abundant universals like this one in order to be able to account for the connection between Ts and Os in depictions of particulars, without the aid of additional means of identification. Additional identification is unavailable on a perceptual account, since this type of account involves that seeing T in a picture is to see a picture of O. Thus we cannot fix the connection between Ts and Os through, for instance and as Zeimbekis (2010) suggests, our communication about the picture as a historical object. It must simply be in virtue of seeing T in the picture that the picture depicts O (although what makes it the case that T is seen may depend on a historical link). Now, with regard to pictures of general things, a universal is sufficient for the connection to the depicted object. For instance, a picture presenting us with the universal of being a man determines a man but no particular man as the depicted object. My idea is that also for pictures of particulars a universal can be sufficient for the connection to the depicted object, since the universal of being the Houses of Parliament is instantiated only by the Houses of Parliament. In this way, the proposal that Ts are universals looks promising because it can make sense of the connection between Ts and Os both for pictures of something general and for pictures of something particular.

Let us therefore explore further how this proposal may be developed. Let us start by considering the following challenge for the account. Suppose we have a picture that depicts a woman but no woman in particular. According to the present version of the perceptual account, the universal of being a woman is then present and seen in the picture. But the picture does not instantiate the universal of being a woman; the picture is not a woman after all. In what sense, then, is the universal present in the picture, given that it is not instantiated by the picture?

In answering this question, I borrow an idea from Wiesing. Recall that one of the constraints we recovered from his discussion is that Ts are ‘artificially present’, i.e. merely visibly present, in pictures. As I mentioned, this idea need not imply that Ts are pure visibilia, as Wiesing himself holds. Rather, we may simply take the idea to be that sometimes, and typically in pictures, things can be present to only one of our senses, namely the sense of sight. This, I suggest, is the case with the presence of universals in pictures.

Depending on how one conceives of universals, this idea that universals are merely visibly present in pictures may be developed in different ways. Let us first consider how it plays out on a Platonic conception of universals, according to which universals are abstract entities that do not have a spatio-temporal location. Seeing a universal in a picture will then, according to this version of the perceptual account, amount to seeing into a realm outside space and time; it amounts to seeing into Plato’s third realm. This makes presence of a universal in a picture and instantiation come clearly apart: Since there can be uninstantiated universals on the Platonic conception, we can see universals in pictures that only exist in the third realm and that are not instantiated by any objects, including the picture itself.

The distinction between presence in pictures and instantiation seems more problematic to make on Armstrong’s (1989) and Lewis’s (1986) conception of universals, according to which universals have spatio-temporal location and do not linger in a third realm. This type of conception often involves that universals are wholly present in their instances, i.e. that they are present throughout their instances and not only in one spatio-temporal part of them. How can the presence of a universal in a picture differ from its presence in its instances, given this conception of universals?

My preferred answer is that only the presence of a universal in a picture facilitates visually perceiving the universal. This answer, however, brings with it a specific commitment concerning perception, namely that we generally perceive material objects themselves and not their universals or properties (except for when an object is experienced as a representation).Footnote 11 I cannot defend this commitment here. But I can defend the claim that universals are perceived in pictures but not in non-representational objects. This can be defended by arguing, more generally, that we seem to experience a different type of thing in pictures than in face to face experiences of non-representational material objects. Let me mount a defence by appeal to a couple of examples.

Consider a case of portraiture, such as Gullvåg’s painting of the current Norwegian Queen, Sonja.Footnote 12 In this portrait, the artist arguably succeeds in capturing Queen Sonja’s personality.Footnote 13 When experiencing a person himself or herself, by contrast, we arguably see the person, not the person’s personality. We may, admittedly, be able to see a personality through paying attention to, e.g., the lines of a person’s face, as the person’s moods and experiences have shaped them, and this can be done when the person, rather than his or her portrait, is before us. But contemplating a person thus, I think, counts as treating the person as an artwork. One is not simply seeing the person, but rather contemplating his or her appearance and thereby seeing something else, namely his or her personality. This seems similar to how we in contemplating a picture thereby see something in the picture. So, insofar as we can see personalities in pictures, there is something we see in pictures that is not seen in experience of non-representational objects.

Support of the claim that we see a special type of thing in pictures is also provided by a phenomenon Lopes (2005) labels ‘inflection’. This is the phenomenon that what Lopes calls the ‘design’ of the picture, i.e. the visible surface properties in virtue of which a picture depicts, may transform our experience of what is seen in the picture. This may lead to us seeing things in pictures that we cannot see face to face. For instance, Lopes provides the following example to illustrate. In Honoré Daumier’s drawing entitled Fatherly Discipline,Footnote 14 the face of the disobedient child is drawn by using sharp V-shaped pen strokes that contribute to the expression on the child’s face. As a result, Lopes writes, ‘[w]hat the face looks to express depends on the design in a way that has no analogue in natural expression’ (ibid., p. 79). In this way, he thinks, ‘a depicted expression may look other than it would look when seen with the naked eye’ (ibid.).

Finally, another phenomenon that renders support to the claim that the type of things seen in pictures can differ from the type of things we see when experiencing non-representational objects is the phenomenon of indeterminacy. An extreme case exhibiting indeterminacy is a stick figure picture of a person. Such a picture is indeterminate with respect to whether what is seen in the picture is something male or female, tall or short, skinny or fat, etc. By contrast, we do not see something indeterminate in this way face to face. While we may see a person in a dimly lit street and not be able to tell much about his or her height, build, etc., it remains a fact that the person we see has a particular height, build, etc., and hence we do not see something indeterminate. Now, there is a question as to whether indeterminacy in pictures pertains to just to the depicted object, O, or also to what is seen and present in the picture, T.Footnote 15 If it only pertains to O, difficult questions arise concerning the connection between the determinate T and the indeterminate O. Thus, what seems most straightforward on the present version of the perceptual account is to hold that both T and O are indeterminate with regard to, e.g., height, build, etc. In the stick figure picture, T is the universal of being a person, and O is a person but no person in particular. This makes the connection between T and O very clear; it is the connection between a universal and what instantiates it.

Let us take stock. The examples provided do not establish that pictures always display things that are not seen face to face when presented with non-representational material objects, but they make the idea more plausible. If the idea is accepted, we can bypass the abovementioned challenge of accounting for the difference between instantiation and the presence of universals in pictures, also on Armstrong’s and Lewis’s conception of universals. Hence, the version of the perceptual account according to which Ts are universals can be developed by relying on either the Platonic or the Armstrong-Lewis conception of universals. Relying on the Platonic conception may have an advantage in that it would allow for a straightforward extension of the perceptual account to pictures of fictional entities, since there can be uninstantiated Platonic universals. Relying on Armstrong’s and Lewis’s conception, by contrast, would be preferable to the extent that this conception on independent grounds is more widely accepted and more easily defensible than the Platonic conception. For the purpose of this paper, however, I do not make a choice between these two developments of the account.

6 Conclusion

By way of conclusion, it is appropriate to make one final observation about the difference between construing Ts as universals, as visual appearances, and as material objects. In the previous section, I explained that in order to avoid the connection problem and account for the relationship between Ts and Os with regard to pictures of particular objects we need to acknowledge universals like that of being the Houses of Parliament. Note that an analogous move can be made also if we construe Ts as appearances or as merely possible material objects. If Ts are appearances, we need only acknowledge appearances such as being identically-looking to the Houses of Parliament in order to avoid the connection problem. Similarly, if Ts are merely possible material objects, we need only acknowledge that such objects can have the property of, e.g., being identical to the actual Houses of Parliament. However, to the extent that such looks and such modal properties are more contentious postulations than that of universals like the universal of being the Houses of Parliament, construing Ts as universals is preferable.

Alternatively, a good approach may be to not regard the different choices of Ts discussed above as mutually exclusive. In my view, a plausible version of the perceptual account is one that admits of universals, appearances, material objects, and perhaps also other types of entities in addition as Ts. We may even admit that several types of T can be present and seen in the same picture. While such a version of the perceptual account may raise difficult metaphysical questions concerning if, and how, the different types of Ts relate, it also allows for advantageously adapting the account to the special features of each picture, thus displaying sensitivity to the large variety of pictures that there is.