SEEING-IN AND SINGLING OUT: HOW TO RECONCILE PICTURES WITH SINGULAR THOUGHT

: According to the standard view of pictorial reference, a picture pro-duces singular thought in virtue of both its appearance and its history. Zeimbekis (2010) challenges this view, arguing that the perception of the picture ’ s appearance does not contribute to the production of singular thought. The paper de-fends the standard view from Zeimbekis ’ challenge, specifying the roles of appearance and history in pictorial reference. While knowledge about the picture ’ s history allows one to identify the standpoint from which to see the scene depicted, the perception of the picture ’ s appearance exploits that standpoint to single out individuals in that scene.

form a singular thought about that boat in virtue of seeing it in the picture. 1 Philosophers have addressed this issue in the framework of the debate on pictorial reference. The notions of reference and singular thought are strictly connected. That is because one can conceive of reference as a four-place relation such that a subject uses a sign to elicit a singular thought about an individual from an audience (Bach 1994;Strawson 1950). From this perspective, reference is the relation whereby signs elicit singular thoughts.
If the sign at stake is a picture, reference to the things depicted can be called pictorial reference. Philosophers tend to agree on the claim that the picture's appearance alone cannot secure pictorial reference to a particular individual, because several distinct individuals can share that appearance. Thus, the picture's history should be taken into account. But which roles do history and appearance exactly play in determining pictorial reference? Here is where disagreement arises.
According to Goodman (1968), the picture's history determines pictorial reference independently of the picture's visual appearance, just as the reference of a name is determined independently of that name's phonetic appearance. 2 Schier (1986), instead, argues that both history and appearance play a crucial role in fixing pictorial reference. The picture's history warrants reference by connecting us to the things depicted, while the picture's appearance makes this reference pictorial by enabling us to single out the things depicted in a specific way, that is, by deploying 'specific recognitional abilities' (Schier 1986, p. 92).
Schier's account is developed by Lopes (1996), who supplements it by relying on Evans' (1982) notion of demonstrative identification. The latter is the act whereby one makes us think of the relevant individual on the basis of a perceptual link to her. Lopes argues that pictures enable demonstrative identification in the same way as terms such as 'this' or 'that' do. Just as one can refer to an individual by pointing at her and saying 'this', one can refer to an individual by showing a picture of her. From this perspective, the picture's appearance is as crucial as the picture's history to pictorial reference. 3 Zeimbekis (2010) challenges Lopes' account by arguing that picture perception, unlike ordinary perception, does not satisfy Evans' requirements for demonstrative identification. Therefore, the picture's appearance does not contribute to enabling singular thoughts about the things depicted. 4 For Zeimbekis, indeed, the perception of the picture's appearance limits itself to descriptively enriching a singular thought that is independently produced by knowledge about the picture's history. This leads him to refuse the 'dual view' advocated by Schier and Lopes according to which both history and appearance contribute to pictorial reference, thereby going back to Goodman's claim that the picture's history determines the picture's reference independently of the picture's appearance.
In this paper, I aim to resist this conclusion and to propose an amended version of the dual view of pictorial reference. Let us suppose, for the sake of the argument, that Zeimbekis succeeds in showing that picture perception does not satisfy Evans' requirements for demonstrative identification. Nevertheless, this statement does not entail that pictures cannot enable singular thoughts about the things depicted in virtue of depicting them. There might be a variety of singular thought that is distinct from demonstrative identification, and specific to picture perception. I will argue that there is such variety.
Specifically, in Section 1, I will present Zeimbekis's challenge to Lopes's account of pictorial reference. In Section 2, I will figure out the ontological premises on which Zeimbekis's account of pictures and singular thought relies. In Section 3, I will show that such an ontology does not involve that picture perception cannot enable singular thoughts about the things depicted; rather, it involves that picture perception can do so in a specific way, which is different from demonstrative identification. In Section 4, I will elucidate this specific way by comparing and contrasting pictures with names. In Section 5, I will draw my conclusion.

Zeimbekis' claim: pictorial reference without picture perception
The argument whereby Zeimbekis challenges Lopes' account of pictorial reference relies on a Main Premise and two requirements, namely, the Relational Requirement (RR) and the Epistemic Requirement (ER). Ordinary perception provides us with a paradigmatic case of singular thought by satisfying both (RR) and (ER). When a subject S sees an object O, there is a causal connection between O and S, and therefore, (RR) is satisfied. Furthermore, ordinary perception allows S to discriminate O from all other things, because it allows S to situate O in a spatiotemporal location that is unique to it. Therefore, (ER) also is satisfied.

Main Premise
For the sake of the argument, Zeimbekis assumes that pictures can satisfy (RR). He does so by focusing on the case of photographs, which normally warrant a causal connection between the viewers and the objects photographed. Yet he challenges Walton's (1984) and Lopes' (1996) claim that photographs, as such, can enable singular thoughts about the things depicted in virtue of depicting them. He states that even if photographs can satisfy (RR), they cannot satisfy (ER) because they 'do not convey information about the location of their causal relata.' (Zeimbekis 2010, p. 13).
From this perspective, picture perception differs not only from ordinary perception but also from mirror-based perception. In order to show this, Zeimbekis points out that ordinary perception has two basic features that support (ER), namely, Objective Coincidence and Domain Restriction.
Objective Coincidence is 'a coincidence of the egocentrically represented locations of particulars with their objective locations.' (Zeimbekis 2010, p. 13).
Domain Restriction is a grasp of the spatiotemporal domain in which our perception occurs.
In ordinary perception, we represent things as being exactly where they actually are; hence, Objective Coincidence is satisfied. Moreover, we know that what we are perceiving is occurring here and now, and thus, 'the domain itself is picked out from other domains by perception' (Zeimbekis 2010, p. 13); hence, Domain Restriction also is satisfied.
While ordinary perception has both Objective Coincidence and Domain Restriction, mirror perception usually has the latter but lacks the former. In watching a mirror, we see things in front of us, whereas they are in fact in face of the mirror, and therefore, there is no Objective Coincidence. Nevertheless, there is Domain Restriction, because we know that the scene we see is occurring here and now, although not in the location where we are seeing it. According to Zeimbekis, Domain Restriction alone is sufficient to warrant (ER): if we see something in a mirror, we perceive its appearance, and we know where and when it is occurring, and therefore, we are capable of discriminating it from all other things.
Photographs, unlike mirrors, lack not only Objective Coincidence but also Domain Restriction. A mirror tells us that the scene we are seeing is occurring here and now, in face of the mirror, whereas a photograph does not tell us anything about where and when the scene depicted occurred. Therefore, photographs do not satisfy (ER): we cannot discriminate something we see in a photograph from all other things.
Consider for example the photograph of a donkey D 1 . There may be another donkey D 2 that has exactly the same appearance as D 1 . In ordinary perception, and even in mirror-based perception, this would not affect (ER) because we could discriminate D 1 from D 2 by relying on its unique spatiotemporal location. Yet, in picture perception, we have no means to discriminate D 1 from D 2 by relying on locations because such perception does not provide us with information about the locations of the objects perceived. 6 This leads Zeimbekis to conclude that picture perception cannot acquaint us with the objects depicted, and therefore, given the Main Premise, pictures cannot enable singular thoughts about the things depicted in virtue of depicting them. Picture perception can only produce descriptive thoughts, which require some cognitive supplementation in order to be turned into singular thoughts. 7 All we can know by looking at a picture is that 'there is (or, if the Relational Requirement is not satisfied, there may be) something whose appearance is like this'. In order to entertain a singular thought about that something, we should look outside picture perception.

The ontology of pictures
Zeimbekis does not deny that pictures can enable singular thoughts about the things they depict. What he denies is just that pictures do so in virtue of their appearance. Yet, if one statesas he doesthat pictures can enable singular thoughts, there must more be in a picture than its appearance. That is to say that, from an ontological point of view, a picture is an object whose identity also depends on its history. As Zeimbekis (2010, p. 11, my emphasis) puts it: the contents of picture perceptions do not themselves provide the kind of numerical and contextual information required for singular thought. Picture reference is instead secured by independent beliefs or linguistic communication about the causal history of pictures as objects. In other words, it is beliefs about the numerical identity of pictures as objects that anchor the reference of the representational contents of pictures.
Such ontology of pictures can be exemplified by considering the case of a photograph of one of two twins. In this case, as Zeimbekis (2010, p. 17, my emphasis) writes: what anchors the representational contents of the photograph to a particular in thought is the perception of the photograph as an object, not its perception as a representation. If I did not have knowledge of the picture's causal history, I would not know which mental file (that for twin A or that for twin B) to feed the perceptual-style information into.
That picture is what it is, namely a picture of A, in virtue of its causal history that relates it to A. From this perspective, the identity of a picture depends not only on its appearance but also on its history. In Wollheim's (1980) terms, the identity of a picture depends not only on its capacity to elicit the peculiar perceptual experience that he calls 'seeing-in' but also on a norm that he calls 'standard of correctness.' The latter, which is grounded in the picture's history, specifies the correct way in which the suitable viewer should experience the picture: 'what the standard does it to select the correct perception of a representation out of possible perceptions of it' (Wollheim 1980, p. 137).
However, I argue, such ontology of depiction does not necessarily lead us to the conclusion that pictures cannot enable singular thoughts about the things depicted in virtue of depicting them. On the one hand, I agree with Zeimbekis that a picture, as an object, is constituted by two components, namely, in Wollheim's terms, a visual appearance which supports a seeing-in experience, and a standard of correctness which supplements that experience. On the other hand, I challenge Zeimbekis' claim that seeing-in only supplies descriptive information, so that singular thoughts about the things depicted are enabled exclusively by the standard of correctness. 8 The Wollheimian ontology of pictures does not forcefully lead us to Zeimbekis's thesis on pictures and singular thought. There is another alternative at our disposal: the standard of correctness and the seeing-in experience can cooperate to produce singular thoughts. Specifically, the standard can provide seeing-in with Domain Restriction (i.e., the individuation of the relevant spatiotemporal domain), which allows that experience to single out the individuals depicted. That is to say that the standard does not produce singular thought on its own but rather sets up the conditions whereby a seeing-in experience can single out the things depicted. Resorting to a sport metaphor, the standard just makes an assist, but it is picture perception that makes the point, that is, out of metaphor, the reference.
The standard of correctness is thus crucial to provide Domain Restriction for pictures such as photographs or portraits that are meant to produce singular thoughts about individuals located at certain places and times. Yet, for Wollheim (1980), the standard of correctness does a lot more than providing Domain Restriction (Terrone 2021). The standard, for instance, plays a crucial role in pictures such as Sacra Conversazione paintings which portray certain subjects but are not meant to locate them at certain places and times. 9 In the latter case, reference, if any, should be fixed directly by the standard of correctness, just as Zeimbekis states. Yet, Zeimbekis contends that this holds true for any picture, while I am arguing that this does not hold true when pictures have a standard of correctness that fixes Domain Restriction. In that case, picture perception contributes to pictorial reference. This is the thesis on pictures and singular thought that I will defend in what follows.

Challenging Zeimbekis' claim: how picture perception contributes to pictorial reference
According to Zeimbekis, picture perception cannot enable singular thoughts about the things depicted because it cannot satisfy the Epistemic Requirement. Picture perception cannot do so because it lacks Domain Restriction, that is, a grasp of the spatiotemporal domain in which what is perceived occur. Yet Zeimbekis overlooks the possibility that picture perception can be provided with Domain Restriction by the standard of correctness. We can acknowledge this possibility if we consider that picture perception shares with ordinary perception a representational structure that represents a certain scene from a standpoint S at a time T. 10 The difference between picture perception and ordinary perception lies in the values of the variables S and T. On the one hand, in ordinary perception, S has the value 'here' (where my body is), and T has the value 'now' (when my experience is occurring). On the other hand, in picture perception, S and T seem to remain indeterminate. Yet this would be the case only if we treated a picture as constituted by nothing but its visual appearance. Instead, according to the Wollheimian ontology of pictures sketched above, a picture is constituted by both its visual appearance and its standard of correctness. I argue that both the appearance and the standard contribute to picture perception. The appearance elicits an experience that represents the scene depicted from a standpoint S and at a time T while the standard can fix the values of S and T. For instance, in the case of a photograph, S has the value 'where the camera was' and T has the value 'when the photograph was taken'. In this way, the standard of correctness provides picture perception with the Domain Restriction that allows the viewer to perceptually single out the individuals depicted.
I agree with Zeimbekis that there is an important difference between ordinary perception and picture perception as regards singular thought. Yet, I argue, the difference is not as sharp as he states. The difference is just that ordinary perception fixes the values of S and T on its own -S has the value 'here' and T has the value 'now'whereas picture perception cannot do so. Yet this difference does not entail that picture perception cannot enable singular thoughts about the things depicted and thus, only supplies descriptive information to singular thoughts that are produced independently of it. This difference only entails that picture perception, unlike ordinary perception, cannot enable singular thoughts on its own. Picture perception needs the help of the standard of correctness to so, but the fact that one carries out a task with the help of someone else does not mean that one does not carry out that task! Interestingly, Zeimbekis, in his criticism of Walton's account of photography, accepts that mirror perception can produce singular thoughts with the help of some supplementary knowledge: Walton holds that if a series of mirrors, the number and arrangement of which is unknown to the viewer, causes a reflection of a carnation, then the subject will have no knowledge of the carnation's location, but will nevertheless count as seeing it. Here, everything depends on how the example is set up in certain respects which remain underspecified in Walton's example. For example, if the subject knows how much space is covered by the mirroring system and knows that that space contains only one carnation, then the example turns out to be irrelevant; then mirror perception will still cause a singular thought in a way that photographs cannot, namely, by preserving domain restriction. (2010, p. 14, my emphasis) In this example, it is clear that mirror perception is supplemented by some knowledge that allows it to enable singular thoughts about the things mirrored. One might call the content of this knowledge 'the standard of correctness of the mirror,' that is, the rule that governs its use. Why does Zeimbekis allow mirror perception to exploit supplementary knowledge in order to enable singular thoughts but prevent picture perception from doing so? This seems to be an ad hoc move.
The following passage of Zeimbekis' paper is quite significant in this respect: Suppose that I enter an unfamiliar surveillance room in which live images from several identical stairwells appear on different screens. Each time I look at one of the screens, several locations satisfy my visual contents; any thought I formulate about a stairwell on the basis of my perceptual contents will have the cognitive role of a de dicto thought, albeit one with highly determinate descriptive content. (2010, p. 14) If one allows the viewer to exploit supplementary knowledge in the case of mirror perception, as Zeimbekis does, why do not allow the viewer to do so also in the case of live video coverage? Even if the different monitors portray what is going on in 'several identical stairwells,' if I know the standpoint S 1 corresponding to monitor M 1 , the standpoint S 2 corresponding to monitor M 2 , and so forth, I can perceptually single out the individuals displayed on the different monitors just as I do in the case of mirror perception. I can see no reason why one can exploit supplementary knowledge to obtain Domain Restriction in the case of mirrors but not in the case of live video surveillance. This shows that at least some kinds of pictures can elicit singular thought just as mirrors do.
What holds true for live video surveillance also applies to live television broadcast. For instance, when one watches a tennis match on television, picture perception can contribute to enable singular thoughts about the two players. That is because the standard of correctness of this moving picture provides us with the information that the event we are seeing is occurring at a certain time (say, now, if this is a live broadcast) and in a certain place (say, Wimbledon Centre Court). Even if two indiscernible Twin Players are playing an indiscernible tennis match on Twin Earth, the standard of correctness allows us to discriminate the players we are seeing from their Twin Earth twins (just as from all other things) thereby entertaining singular thoughts about them. And we do so within, not without, picture perception. Just as in the case of mirrors, singular thought is produced by perception itself, albeit supplemented by some knowledge.
At this point, one might still object that live video surveillance or live television broadcast is not paradigmatic cases of pictures. Most picturesso the objection runsare such that their viewers usually ignore the standpoint S from which, and the time T at which, a certain scene is depicted. Yet this PACIFIC PHILOSOPHICAL QUARTERLY objection can be contested by noting that ignorance of the exact place and time does not prevent the viewers of a picture to entertain singular thoughts about the individuals depicted. Something similar holds true in ordinary perception, which can enable singular thoughts in spite of ignorance of the exact time and place of the events perceived; knowing that those events occur here and now is quite enough to allow for singular thought. Likewise, what the viewers of a picture should know in order to form singular thoughts is just that there is a unique place S and a unique time T for the picture. Of course, knowing the exact values of S and T can improve our understanding of the picture, as it happens in the case of those digital cameras that supply metadata specifying when and where a picture was taken. Yet even if such data are not available, a descriptive assumption such as that S is 'where the camera was' and T is 'when the picture was taken'the here and now of the camera, as it wereis quite enough to obtain Domain Restriction. 11 In making this assumption, the viewers rely on deference to the relevant practice of picture production.
The possibility of obtaining Domain Restriction independently of picture perception, however, does not lead us back to the Goodmanian claim that pictorial reference is determined independently of picture perception. Domain Restriction, indeed, is only necessary for pictorial reference, not sufficient. Once Domain Restriction is obtained by taking the picture's history into account, picture perception is still needed to single out the relevant individuals in that domain thereby finalizing singular thought. Hence, picture perception crucially contributes to pictorial reference by bridging the gap between mere Domain Restriction and full-fledged singular thoughts about the things depicted. In this sense, pictures enable an experience that one might call an indirect perception of the things depicted. This experience deserves to be called perception inasmuch as it perceptually singles out the things depicted. Yet such perception is indirect because, unlike ordinary perception, it requires the mediation of extra-perceptual information in order to fix the domain in which those things can be perceptually singled out.

Pictures and names
The main target of Zeimbekis' argument is Lopes' claim that the kind of singular thought enabled by pictures can be traced back to demonstrative identification. In fact, demonstrative identification does not require supplementary knowledge in order to obtain Domain Restriction. If I tell you 'look at this', you automatically know the place where and the time when this occurs. Conversely, if I show you a picture you cannot automatically establish the place and the time of the scene depicted. Therefore, according to Zeimbekis, picture perception, unlike demonstrative identification, cannot enable singular thoughts.
I have argued that this conclusion is too hasty. Picture perception might enable singular thoughts in a way different from demonstrative identification. Thus, the fact that pictures do not function as demonstratives does not entail that they make reference in a perception-independent way just as proper names do. Although pictures differ from demonstrative just as names do, there is also a relevant sense in which pictures differ from names.
On the one hand, both pictures and names, unlike demonstratives, do not enable singular thoughts in a straightforward perceptual way but rather rely on a causal-historical chain to do so. On the other hand, pictures differ from names because this chain plays a different role in the production of singular thoughts. In the case of names, the chain leads us directly to the individuals the singular thoughts are about. In the case of pictures, instead, the chain only leads us in the proximity of the relevant individuals. Thus, the causal chain the connects a picture to an individual depicted does not give us the individual itself, but only a standpoint from which, and a time at which, that individual can be singled out. In order to single that out, we need picture perception.
A further comparison between pictures and proper names, within the framework of Evans' (1982) account of reference, on which both Zeimbekis and Lopes rely, can be instructive in this respect. A proper name, say NN, understood as a mere sequence of letters, cannot enable singular thoughts about its bearer B because the sequence of letters NN could be used also to name another individual C, and therefore, in hearing a sentence such as 'NN is at home,' one cannot discriminate B from all other individuals just by considering the appearance of the name.
A proper name can enable singular thought only if it is endowed with a sort of standard of correctness, which selects the correct bearer out of possible bearers. As Evans puts it, 'It may help […] if we think of individuating the words of a language not only phonetically but also by reference to the practices in which they are used. ' (1982, p. 384;Kaplan 1990, p. 111) By 'practice', here, Evans means the practice of production of a name, that is, the introduction of a name in a certain linguistic community that Kripke (1980) calls 'a baptism' (cf. Evans 1982, p. 385). The baptism plays for the name the role that the history of production plays for the picture: they both place a normative constraint, namely, a standard of correctness, upon the uses of a certain cognitive artifact (name or picture). 12 Thus, a proper name can enable singular thoughts only if one takes history into account. This means considering the standard of correctness, that is, the normative constraint that the baptism imposes on the users of the name. The standard is what makes a proper name unique in spite of sharing its appearance with several other names.
Even if one cannot directly discriminate the bearer of a certain name from all other individuals, one can discriminate this very name from all other names by means of its standard of correctness. This involves relying on the practice that has established the standard thereby indirectly discriminating the bearer of the name from all other individuals. In this sense, a proper name functions as 'a cheque for identification that could be cashed by discovering the use of producers' (Evans 1982, p. 392).
Pictures resemble proper names in this respect. Specifically, picture perception differs from ordinary perception just as the use of proper names differs from the use of indexicals. Evans (1982, p. 373) argues that the latter are 'one-off devices' that differ from proper names because they can directly enable singular thoughts whereas names can do so only indirectly, by relying on a practice that originates from a baptism. Likewise, I contend, pictorial perception differs from ordinary perception because the latter can directly enable singular thoughts, whereas the former can do so only indirectly, by relying on a practice that has established a standard of correctness. Of course, this does not show that proper names and pictures cannot enable singular thoughts. Rather, this shows that they enable singular thoughts in a practice-dependent way which is different from the immediate, 'one-off' way of indexicals and ordinary perception.
Specifically, a picture fails to enable singular thoughts about the individuals depicted if the viewer fails to entertain a singular thought about the picture itself, that is, a thought that discriminates this picture from all other pictures. Yet, if the viewer succeeds in entertaining the latter thought, she can also entertain singular thoughts about the individuals depicted. That is because, by picking out the picture for what it is, namely, a visual appearance coupled with a standard of correctness, the viewer can acquire Domain Restriction, which allows her to perceptually single out the individuals depicted.
Here, we come to the crucial difference between pictures and names. They both can be conceived of, borrowing Evans' metaphor, as cheques for identification that could be cashed by discovering the use of producers. Yet what one gets when the cheque is cashed is different in the two cases. Cashing the name-cheque gives us the relevant individual, namely, the bearer of the name, whereas cashing the picture-cheque only gives us a standpoint from which and a time at which we can perceptually single out the individual depicted.
This difference between pictures and names effectively explains why cashing the name-cheque only allows us to single out one individual whereas cashing the picture-cheque might allow us to single out several individuals, that is, all the individuals that the picture shows us from a given standpoint and at a given time. Consider for example Kent State Shootings, the picture taken at Kent State University on 4 May 1970 for which John Paul Filo won a Pulitzer prize in 1971. 13 The perception of this picture enables singular thoughts about several individuals. In forming such thoughts, one surely relies on the standards of correctness, which tells one that the relevant standpoint is located at Kent State University on 4 May 1970. Yet this piece of information, as such, cannot produce singular thoughts about the individuals depicted. It only provides one with the Domain Restriction that allows one to perceptually produce those singular thoughts by scrutinizing the picture. 14 If one hears a name during a conversation, one can form a singular thought about the bearer of that name independently of the name's appearance. One just needs to deploy a mental file or 'dossier' about 'the dominant source' of the informational chain that has led the speaker to utter that name (cf. Evans 1973, p. 200). 15 Yet, when one faces a picture like Kent State Shootings, deploying a mental file about the dominant source of information is not sufficient. That source only provides us with a perspective that we should exploit to perceptually single out the individuals depicted thereby forming singular thoughts about them.
Ultimately, in the case of a proper name, history is sufficient to enable singular thoughts. Appearance does not matter because the hearer should just deploy a mental file about the dominant source of information of that utterance of the name. If one hears somebody uttering the name 'Mary Ann', one can form a singular thought about the dominant source of information of that utterance of 'Mary Ann' without the need of scrutinizing the auditory appearance. Whatever auditory appearance might replace that of 'Mary Ann' if properly connected to the dominant source of information. Yet, when one forms singular thoughts about the individuals depicted by a picture such as Kent State Shootings, the picture's appearance plays a crucial role because one should scrutinize it in order to form those thoughts. Thus, when it comes to singular thought, the visual appearance of a picture could not be replaced by whatever appearance in the way the auditory appearance of a name might be replaced.
A special case in which one can just rely on the dominant source of information is that of visual traces such as fingerprints or footprints. If one sees, for instance, a footprint on the sand, one can form a singular thought about its dominant source of information, that is, the individual who caused it, without the need of scrutinizing it. Yet pictures are different for footprints in this respect, at least if we endorse Wollheim's claim that a picture generates a seeing-in experience such that we 'discern something standing out in front of, or (in certain cases) receding behind, something else. ' (1987, p. 46, my emphasis) By generating this kind of experience, a picture allows us to form singular thoughts about at least two things, namely, what Wollheim calls the 'something' and the 'something else.' In order to form these thoughts, we need to visually distinguish the 'something' from the 'something else,' and thus, we need to take not only the picture's history but also its appearance into account. Singling out the 'something' and the 'something else,' indeed, requires not only considering the causal-historical information that leads us in the proximity of those two things but also perceiving the appearance that allows us to distinguish one thing from the other.

Conclusion
According to Lopes, pictures enable singular thoughts just as demonstratives do whereas, according to Zeimbekis, pictures enable singular thoughts just as proper names do. I have argued that they are both right and wrong in different respects. Lopes is right in stating that pictorial reference has a crucial perceptual component, but he is wrong in assimilating this component to that at stake in demonstrative identification. Zeimbekis is right in pointing out that pictorial reference differs from demonstrative identification because it lacks Domain Restriction, but he is wrong in inferring the conclusion that pictures enable singular thoughts in the same way in which proper names do so. The right conclusion is more articulated. On the one hand, a picture functions like a proper name with respect to reference as far as they both depend on informational chains. On the other hand, the informational chains of names directly single out the relevant individuals, whereas the informational chains of pictures only fix the domain of reference thereby allowing the viewer to perceptually single out the relevant individuals. Although seeing-in, unlike ordinary perception, cannot single out the things depicted on its own, it crucially contributes to the formation of singular thoughts about those things. This shows that pictorial reference is neither an instance of demonstrative reference, as Lopes argues, nor a variant of name reference, as Zeimbekis argues. Rather, I have argued, pictures involve a variety of reference which has its own specificity. 16 Dipartimento di Antichità, Filosofia, Storia (DAFIST) Università degli Studi di Genova, Genoa, Italy NOTES