The Epistemic Status of Photographs and Paintings: A Response to Cohen and Meskin Dan Cavedon-Taylor Forthcoming (2009) in The Journal of Aesthetics and Art Criticism (vol. 67, no. 2). Please refer to published version only. Why do photographs occupy an epistemically privileged status amongst types of depictions? As Jonathan Cohen and Aaron Meskin put it: “Why are photographs epistemically special in a way that other sorts of depictive representations are not? (Why, for example, do photographs but not paintings carry evidentiary weight?)”1 Orthodoxy on this matter has it that the answer has to do with the nature of the causal relation that photographs bear to their subjects. In a series of articles, Cohen and Meskin have proposed a rich and novel answer to the above question that, as I read it, bucks this orthodoxy. Cohen and Meskin’s account of photography’s epistemic value, in addition to referencing the nature of the relation that photographs bear to their subjects, references viewers’ background beliefs about photographs as well.2 Whilst this gives their account of photography’s epistemic value a degree of sophistication and nuance that other ones may lack, it seems to me that what Cohen and Meskin have added thereby renders their account internally unstable. I. According to Cohen and Meskin, photographs are epistemically special because: 1 (1) token photographs are spatially agnostic sources of information, and (2) viewers hold background beliefs about the category of photographs that influence their attitudes towards the epistemic status of viewed token photographs.3 This is a lot to unpack and (1) further decomposes into two claims. The first of these claims in (1) is that photographs are epistemically valuable because they transmit information about visually accessible properties of their subjects, e.g., colour and shape properties, what Cohen and Meskin call “v-information.”4 However, photographs are alleged by Cohen and Meskin to transmit vinformation in a manner independent of their transmitting egocentric spatial information (i.e. information about where, relative to the viewer’s position in space, the photographed object is or was spatially located). Meksin and Cohen argue persuasively that photographs fail to transmit such information. This is a second, further claim in (1). That is to say, Cohen and Meskin hold that photographs are not simply epistemically valuable because such pictures transmit v-information; rather, they hold that photographs are epistemically valuable precisely because of the spatially agnostic—i.e. non-egocentric— manner by which this is achieved. In fleshing out this point they argue that when it comes to other sources of v-information (specifically ordinary perception) one only receives v-information insofar one receives egocentric spatial information.5 But photographs give us v-information in a manner that is independent, and 2 therefore not hostage to the fortunes, of egocentric information. Photographs, Cohen and Meskin write, are thus “a relatively undemanding source of information about the visually accessible properties of objects—one that works even when we lack information about egocentric location.”6 This explains (1) above. But Cohen and Meskin think that photographs being spatially agnostic sources of v-information cannot be the whole story when it comes to explaining their epistemic status. Cohen and Meskin believe that types of paintings and drawings, ‘veridical landscapes’ for instance, are spatially agnostic sources of v-information too. Being veridical, this type of hand-made picture is also a source of v-information. Moreover, and again like photographs, hand-made pictures fail to transmit egocentric spatial information, so they too are spatially agnostic sources of v-information. This suggests that paintings and photographs may be epistemically on par. Yet common-sense would have it that photographs are more epistemically valuable than paintings. So photography’s status as a spatially agnostic source of v-information cannot be the whole story when it comes to explaining its epistemically privileged status in relation to painting. Cohen and Meskin’s solution here is to introduce (2). They firstly claim that “the type of photographs is salient for subjects in a way that the type of veridical landscape paintings is not.”7 By this they mean that upon encountering a photograph, viewers typically classify it as such; as ‘a photograph’. But when viewers encounter a veridical landscape painting, say, Cohen and Meskin believe viewers tend not to classify it as ‘a veridical landscape painting’, but only as ‘a 3 painting’. Moreover, Cohen and Meskin hold that viewers typically possess the following background beliefs about these two depictive categories, ‘photographs’ and ‘paintings’: By and large, viewers believe that the type of photographs is one whose members carry v-information. And by and large, viewers believe that the categories to which they assign veridical paintings are ones whose members may fail to carry v-information.8 All this, according to Cohen and Meskin, explains what is epistemically special about photographs. Not only are such pictures a relatively undemanding source of information about visually accessible properties of their subjects, but they are salient in the minds of viewers in a way that paintings are not. Thus, on Cohen and Meskin’s account, the epistemic value of photographs resides in both the nature of the relation such pictures bear to their subjects—i.e. as spatially agnostic information channels—as well as the nature of viewers’ beliefs about the photographic medium in comparison with paintings and other hand-made pictures. II. It seems to me that Cohen and Meskin are guilty of conflating two incompatible conceptions of epistemic value. 4 To begin, note that on (1) the epistemic value of photography looks as if it is viewer-independent feature. (1) is concerned with the nature of the relation that photographs bear to reality; they are alleged to be spatially agnostic sources of vinformation. Now Cohen and Meskin characterise information as being transmitted between token events independent of anyone’s doxastic attitudes.9 A consequence of this is that photographs are spatially agnostic sources of vinformation about their subjects independent of viewers’ beliefs about photographs; that is, whether or not viewers recognise or are sensitive to photographs being spatially agnostic sources of v-information. All this indicates that something of a realist, or viewer-independent, conception of epistemic value is in play here. That is, photographs have epistemic value on (1) not because of viewers’ beliefs about or reactions to photographs, and not because of the epistemic practices photographs are bound up in, rather photographs have epistemic value because they are information channels that transmit vinformation without simultaneously transmitting egocentric information. Photographs are epistemically valuable on (1), even if viewers’ failed to recognise that such pictures are spatially agnostic sources of v-information. The other half of Cohen and Meskin’s account of photography’s epistemic value, (2), attempts to supplement the above by appealing to facts about viewers’ psychological states. However, with a realist or viewer-independent conception of epistemic value in play it is particularly unclear how they can hope to achieve this. It seems to me that the only way in which facts about human psychology, namely the way viewers classify depictive types as well as their background 5 beliefs, can be at all relevant for the epistemic value of photographs is if one takes such value to be viewer-dependent. The problem is that it is not obvious how one can hope to appeal to psychological facts about viewers when attempting to explain to epistemic value of something else—photographs—if one holds, as (1) suggests Cohen and Meskin do, that the epistemic value of photography is independent of viewers’ attitudes. Unless one holds that the epistemic value of photography is something that viewers confer upon the medium—unless one assumes the epistemic value of photography is viewerdependent—then on the face of it it’s not clear why one should be at all interested in viewers’ background beliefs here. For psychological facts about viewers, such as those implicated in (2), cannot be relevant for the epistemic value of photography without assuming a conception of epistemic value that conflicts with the conception that seems implicated by (1). The problem here for Cohen and Meskin is that what makes photographs epistemically valuable cannot be on the one hand, as (1) would suggest, some viewer-independent feature they possess, but simultaneously be a viewerdependent feature on the other, as is suggested by (2).10 Cohen and Meskin’s account of photography’s epistemic value would thus appear to be internally unstable. I should stress that none of this is to cast doubt upon Cohen and Meskin’s claim that photographs are spatially agnostic sources of v-information. Nor do I question their general contention that viewers’ psychological states will be relevant when it comes to explaining why viewers treat photographs differently to 6 paintings. The point is that the conjunct of these two claims doesn’t appear to coherently add up to a viable account of what is epistemically valuable about photographs. For (1) and (2) each assume differing, mutually exclusive conceptions of epistemic value. We have seen that Cohen and Meskin rightly hold that (1) alone isn’t up to the task of explaining what’s epistemically distinctive about photographs. Is (2) alone up to this task? That is, could the epistemic value of photographs be located in viewers’ background beliefs about photographs alone; could such value be purely viewer-dependent? Such an account would take us away from Cohen and Meskin’s project, but it is one worth considering nonetheless. The problem for this view, however, is that, when it comes to photographs at least, epistemic value doesn’t appear to be a viewer-dependent concept. Consider a world whose inhabitants took photographs, but who simultaneously failed to treat such pictures as spatially agnostic sources of v-information. Maybe the inhabitants of the world we are considering value photographs for purely aesthetic reasons alone. Now it seems to me that in such a world photographs have epistemic value, it’s just that the occupants of this world fail to recognise this fact. They are just not sensitive to the epistemic affordances of photographs. The inhabitants of the world we are considering could know various propositions about their environment on the basis of seeing photographs. Despite the beliefs of the inhabitants of this world, photographs are still spatially agnostic sources of v-information; it’s just that photographs fail to be treated as such. The moral is that whether or not agents are sensitive to a source of information being a source 7 of information this cannot change the fact that a source of information remains a source of information. And sources of information are, of course, epistemically valuable. Intuition therefore suggests that viewers’ attitudes towards photographs, such as those implicated in (2), cannot alone go any real distance to explaining the medium’s epistemic value, since epistemic value, when it comes to photographs at least, doesn’t appear to be a viewer-dependent concept. If what I have said above is right, then viewers’ attitudes and responses to photographs—their finding photographs to be objects of epistemic interest or not—does not settle the issue of photography’s epistemic value. Reflection on the above thought experiment suggests that epistemic value attaches to photographs independent of viewers’ responses to them. So in addition to finding that the conception of epistemic value assumed by (2) conflicts with that assumed by (1), we can now also see that, when it comes to photographs at least, the former conception doesn’t appear viable in the first place. (2) alone, like (1) alone, cannot explain what is epistemically special about photographs. III. Perhaps Cohen and Meskin might attempt to salvage their account by replying as follows: Whilst (1) should be understood as providing an account of photography’s epistemic value, (2) should only be understood as explaining photography’s epistemic superiority over paintings and other hand-made pictures. On this view (1) and (2) differ in their explanatory targets and so are no 8 longer in conflict. On the reading now being proposed things stand like this: Photographs have viewer-independent epistemic value—i.e. they are spatially agnostic sources of v-information as suggested by (1)—but since paintings too form types that are spatially agnostic sources of v-information, photography’s epistemic superiority over painting is instead a product of viewers’ background beliefs—a viewer-dependent feature—as (2) suggests.11 Whilst this alternative reading of Cohen and Meskin’s account appears to gel well with some of their remarks in places (indeed, it appears to me that Cohen and Meskin’s claims are somewhat ambiguous between this reading and the reading previously dealt with in II), I believe this formulation of their view to be no less problematic than that which was previously considered. The main problem with this alternative reading of Cohen and Meskin’s claims is that it simply doesn’t seem to yield the correct explanation as to the epistemic divergence between photographs and paintings. The reading currently under consideration seems to prescribe something of quietist account on this matter. It appears to hold that since both photographs and some paintings can be spatially agnostic sources of v-information, we must look beyond the actual nature of these two mediums to explain why the former is of greater epistemic value than the latter, and that the only other place to look to explain this fact is to the psychological states of viewers. If the line of thought presently being entertained is correct, then we are precluded from pointing to any objective, viewer-independent differences between paintings and photographs by way of explaining the two mediums’ divergence in epistemic value. Affirming (2), on this 9 reading, conflicts with providing a realist or viewer-independent account of photography’s epistemic superiority relative to hand-made pictorial spatially agnostic sources of v-information, such as, e.g., veridical landscape paintings. The problem with this alternate reading of Cohen and Meskin’s claims is that reflection suggests there does exist an actual intrinsic viewer-independent difference between paintings and photographs that can explain what is epistemically distinctive about the latter.12 The difference I have in mind consist in how the two media come by their contents and has to do with an issue that Cohen and Meskin, as we saw earlier, claim to be sensitive to; namely, the relative undemanding nature of photography’s epistemic affordances. Specifically, the difference is that far fewer conditions must be satisfied in order for a photograph to be a spatially agnostic source of v-information than must be satisfied in order for a painting to be a spatially agonistic source of v-information. This, it seems to me, is what explains photography’s epistemic superiority relative to paintings, whilst at the same time acknowledging that both can be spatially agnostic sources of v-information. After all, in order for a painting to be veridical, the painter’s visual experience must be veridical. Moreover, the painter must not be inattentive. By contrast, neither of these conditions must be satisfied when it comes to making sure a photograph is veridical. This point is aptly made by Gregory Currie, who writes: If the painter were having an hallucination, and so thinking there was a pink elephant in front of him, his painting would display a pink 10 elephant, not the actual scene before his eyes. With the photograph things are different… It does not matter what the photographer thinks the object in front of the lens looks like; once the camera is set up and film exposed, the camera records the scene actually before it.13 So, against our current reading of Cohen and Meskin’s claims, we can point to viewer-independent differences between paintings and photographs to explain what is epistemically distinctive about the latter, whilst still recognising that both these depictive types may be spatially agnostic sources of v-information. This has been achieved by simply pointing out how many more conditions must be fulfilled by the painter in order that they make their picture veridical, and so be a spatially agnostic source of v-information, than must similarly be fulfilled by the photographer. Painting thus exposes itself at more points than photography to the possibility of misrepresentation and non-veridicality. To the extent that our current reading of Cohen and Meskin’s claims fails to take note of these intrinsic differences between paintings and photographs—differences that seem entirely apt to explain why photographs are epistemically superior relative to paintings— and that they seek, instead, to explain away photography’s epistemically special status relative to painting as simply a product of viewers’ contingent background beliefs alone, this reading too should be rejected. If what I have said here is right, then Cohen and Meskin’s account fails to explain why photographs occupy the epistemically privileged position amongst 11 depictive types that they do. By Cohen and Meskin’s own lights the notion of photographs as spatially agnostic sources of v-information, (1), doesn’t appear up to the job since many paintings, e.g., veridical landscapes, are spatially agnostic sources of v-information too. If viewers background beliefs, (2), are being brought in to supplement (1) in order to explain what is epistemically special about photographs then we have two conflicting conceptions of epistemic value at work and the resulting position is untenable. If, however, as was entertained in III, viewers’ background beliefs, (2), were instead introduced to explain the epistemic superiority of photographs over paintings and not photography’s epistemic value per se, then Cohen and Meskin ignore the widely recognised fact that there are actual intrinsic differences between photographs and paintings with respect to how they come by their contents; differences that appear to unproblematically explain what needs explaining here. Along the way we have seen that the epistemic value of photographs doesn’t look like it can be analysed in purely viewer-dependent terms. Once all the facts about viewers’ beliefs about photographs are in, the question of the medium’s epistemic status still remains open. We are forced to conclude that as rich and sophisticated as Cohen and Meskin’s theory is, it fails to yield the correct account of photography’s epistemic value.14 Dan Cavedon-Taylor School of Philosophy Birkbeck College Malet Street 12 London WC1E 7HX INTERNET: dan.cavedon.taylor@gmail.com 1 Jonathan Cohen and Aaron Meskin, “On the Epistemic Value of Photographs,” The Journal of Aesthetics and Art Criticism 62 (2004), p. 203. 2 See Jonathan Cohen and Aaron Meskin, “On the Epistemic Value of Photographs,” The Journal of Aesthetics and Art Criticism 62 (2004), pp. 197—210 as well as Aaron Meskin and Jonathan Cohen, “Photographs as Evidence,” in Photography and Philosophy, Scott Walden ed. (Blackwell, 2008), pp. 70—91. Purely for reasons of convenience I will, in the main body of this paper, refer to the authors of these papers in the order that their names appear in the first of these articles. 3 4 “Photographs as Evidence,” p. 72. Ibid, p. 74. See also “On the Epistemic Value of Photographs,” p. 204. The conception of information that Cohen and Meskin have in mind here has its roots in Dretske’s Knowledge and the Flow of Information (MIT Press, 1981). 5 6 7 8 9 “Photographs as Evidence,” p. 75. “On the Epistemic Value of Photographs,” p. 204. “Photographs as Evidence,” p. 76. Ibid. See “On the Epistemic Value of Photographs,” p. 201 where they write “Significantly, the claim that x carries information about y is a claim about an objective probabilistic link between the two, and as such its truth is independent of anyone’s doxastic attitudes about the two.” 10 Whether viewer-dependent accounts of some phenomenon deserve to go under the banner of realism or non-realism is a controversial and tricky issue. Even if a viewer-dependent account of x did turn out to be a realist account of x, a viewer-dependent realist account of x would still be in conflict with a viewer-independent realist account of x. 13 11 To be clear, on this reading (2)’s explanatory target has shifted from explaining photography’s epistemic value to now being enlisted to explain why photographs are of greater epistemic import than paintings. 12 Note that arguing for the existence of such differences is not incompatible with viewers having the background beliefs that Cohen and Meskin ascribe them. The point is that, contra how we are currently taking Cohen and Meskin’s claims, such background beliefs wouldn’t then account for why photographs are more epistemically advantageous than paintings. (Indeed, the existence of objective, viewer-independent differences between the two mediums could arguably explain why viewers have the background beliefs about photographs that they do. But it would then be such objective differences themselves, rather than viewers’ background beliefs, which would explain the epistemic privileged status that photographs occupy in relation to paintings.) 13 Gregory Currie, “Photography, Painting and Perception,” Journal of Aesthetics and Art Criticism 49 (1991), p. 24. Many other philosophers also take photography’s being insensitive, in certain respects, to the psychology of the image-maker to be relevant when it comes to the medium’s epistemic status. They include Patrick Maynard, The Engine of Visualization (Cornell University Press, 1997), Scott Walden, “The Objectivity of Photography,” British Journal of Aesthetics 45 (2005), pp. 258—272 and Kendall Walton “Transparent Pictures: On the Nature of Photographic Realism,” Critical Inquiry 11 (1984) pp. 246—277. Thanks to an anonymous referee for advice on this point. 14 Thanks to Stacie Friend, Matthew Rowe and Aaron Meskin for discussion and correspondence on these issues. I am also extremely grateful to the British Society of Aesthetics whose award of their PhD Studentship made this research possible. 14