Critical Inquiry Critical Inquiry

“‘But I am killing them!’ Reply to Palermo and Baetens on Agency and Automatism”

I. Automat, Automatic, Automatism—Again

Critical Inquiry 38:4 brought together a small group of philosophers and art historians, and a leading photographic artist, to assess the roles typically assigned to agency and automatism in understanding recent photographic art. Doing so, we hoped, might advance the theory of photography in ways that disciplinary-specific reflection in either field had not. Charles Palermo, in his wide-ranging response to the ‘Agency and Automatism’ issue as a whole, takes exception to my critique of Rosalind Krauss’s conception of automatism, and what I have to say about that idea more generally.[1] Jan Baetens, in his more narrowly focused reply, takes issue with what I accept on trust from Rosalind Krauss’s account of the photo-novel. Though I do have something to say about both in my original article, I have much more to say about the former: for this reason, and because his reply addresses the central themes of the special issue as a whole, I focus on Palermo’s paper in what follows.

Before addressing their respective criticisms, let me briefly restate the two core claims of my paper. This should put their replies into some kind of perspective. In ‘Automat, Automatic, Automatism’ I set out to answer two key questions: Can Rosalind Krauss get what she needs for her account of artists ‘inventing’ or ‘reinventing’ the medium from Stanley Cavell on the creation of new conventional forms, with which she seeks to underwrite it? Irrespective of whether or not she can get what she needs from Cavell, is her account coherent, taken on its own terms?

Regarding the former, I suggest that she cannot. On Cavell’s understanding of what is required to ‘invent the medium out of itself,’ this could only take place against a background of standing conventions no longer felt to command authority within a given practice. When an art form reaches such an impasse, new conventional forms—what Cavell calls ‘new media’ or ‘automatisms’ within the medium—need to be secured to rekindle such authority and thereby re-establish its community with its audience. On Krauss’s account, by contrast, pervasive scepticism about existing media requires the invention of new media ex nihilo. This voids the grounds of Cavell’s account in advance. So Krauss cannot get what she needs from Cavell. Neither Palermo nor Baetens says anything that makes me think I should reconsider this.

Regarding the latter, I distinguish two senses in which the question of whether artists can invent their own medium may itself be understood: a weaker, descriptive sense that, as things turned out, the (alleged) new medium remained unique to a particular artist, though things might well have turned out differently; and a much stronger, modal sense that the (alleged) new medium could not have been practiced by anyone other than its inventor, given its idiosyncratic nature. Again I say no to both: to my mind, answering yes to either would underplay the constitutive role that publicly debateable norms and conventions of going on, or failing to go on, play in dignifying certain activities but not others as meaningful within a domain. Absent such publicly debateable, historically evolving norms, it is hard to see how what counts as a meaningful extension, or even just a further instance, of the practice in question could be determined—other than by artistic fiat. At that point, the practice ceases to be a practice.

Though I say no to both versions of the question myself, I acknowledge that there is room for debate about this on the weaker, descriptive interpretation of the question that there is not on the stronger, modal interpretation. Krauss equivocates between the two, before eventually coming down in favour of the latter. But the stronger sense evacuates the idea of working in a medium, that is, something separate from the artist that others could in principle work in, of the public dimension required to distinguish it from an artistic variant of solipsism. So Krauss’s account of inventing a medium is not coherent, even taken on its own terms. Neither Palermo nor Baetens takes issue with these claims either, so far as I can see. Assuming that my central claims go through for both, what does that leave?

In my original article, with the themes of the special issue in mind, I focus on Krauss’s claims about the invention of new media in what I call the ‘photographically dependent arts.’ Though, as I point out, nothing I say is meant to hang on this designation and my arguments, if sound, should generalize to Krauss’s similar claims for the inventive powers of artists working in unrelated (non-photographically dependent) art forms. Nonetheless, it is on the terrain of what I have to say about the photographically dependent arts that both Palermo and Baetens chose to make camp: Palermo the better to target my critique of Krauss on William Kentridge, Baetens the better to target my acquiescence to Krauss on James Coleman. Is it just me, or can one really not win in this game?

II.  The Media of Photography

In ‘Reworking or Making Up?’ Baetens focuses exclusively on what I do not question in Krauss’s account of the photo-novel, claiming to find the resulting account ‘frankly absurd.’  Bracing though this is, as the forgoing gloss on ‘Automat, Automatic, Automatism’ should make clear, whether or not Krauss is right about the photo-novel is very much to one side of the core concerns of that paper. Indeed, in so far as my project in that paper was to ask whether Krauss can get what she needs for her account of ‘reinventing the medium’ from Cavell and, if not, whether it is nonetheless coherent setting aside that appeal, I was required to take Krauss’s account of inventing a medium (and with it her account of the photo-novel) as my starting point. Baetens is right to point out that this need not have prevented me from going on to question what Krauss has to say about the photo-novel, its relation or non-relation to the other non-canonical art forms such as comics, or Coleman’s remediation of it in his projected images. But all this strikes me as a paper that Baetens should be writing rather than me, given that he clearly knows far more about this topic than I do. Irrespective of whether Baetens is right to take issue with all this, however, there was no need—given my own concerns—to take a position on any of this myself: indeed in so far as I wanted, after arguing that her appeal to Cavell was unmotivated, to consider her conception of what it takes to invent a medium on its own terms, there was good reason for me not to.

To the extent that I offer any positive account of Coleman at all, it turns not—as Baetens would have it—on Krauss’s account of the photo-novel, but on what Cavell would call the relation between ‘automatisms’ and ‘the automatic’ more generally. That is, the relation between the physical substrate of film as a succession of (automatically produced) still photographic images on a celluloid film strip projected 24 fps onto a screen, and the conventions (or ‘automatisms’) through which this set of merely physical processes and procedures, determined by the laws of chemistry and optics, is animated and thereby rendered artistically significant. To put the same point in Krauss’s terms: my account of Coleman turns on the relation between the works’ ‘technical support,’ slide tape, and the new medium that Krauss believes Coleman’s animation of that support has brought into being, namely ‘projected images.’ What Krauss calls the ‘double-face out,’ and my remarks about this, are just a special case of this more general question concerning the relation between a work’s physical substrate or technical support and the conventions through which this substrate or support is animated, and thereby transfigured, by a given artist. It is this more general question—the question of what Krauss’s theory of artistic media per se commits her to, and not her critical observations about Coleman’s work in particular—that interests me.

This is why my account of Coleman focuses on establishing that the conventions through which he animates his technical support are internally related to that support, in so far as they acknowledge the constraints attendant upon using a still medium for narrative purposes, which might not otherwise be obvious. It is also why my account of Kentridge and Ruscha focuses, similarly, on making explicit the relation between their work’s technical supports and the conventions through which they are animated, on Krauss’s account. As I see it, the constraints imposed by using still images as the basis for a narrative art form brings Coleman’s ‘projected images’ into dialogue with both (staged) photography and (narrative) film, with both of which his projected images share various conventions. Thus, as I note in passing: to the extent that this allows an art form that does not even appear to move to meet Cavell’s characterization of film—by which he has in mind (a subset of) what are typically called ‘movies,’ and not film in general—as a ‘succession of automatic world projections,’ it shows that characterization, as it stands, to be inadequate to the task of isolating movies.[2] For Coleman’s art consists of nothing if not a ‘succession of automatic world projections,’ but not movies. But nor can it be reduced to photography—witness the multi-image dissolves, narrative horizon, voice-overs, and the fact of projection itself—even if it is nonetheless ‘photographically dependent.’

As regards what I do say about Coleman and the photo-novel, which is what Baetens’ reply turns on: I do not claim that photo-novels consist only of double-face outs. It would be very odd if I did: for not only is it patently untrue, but the spread I reproduce to illustrate my own case does not do so! Rather, my reproductions are intended to support the claim that I do make, namely, that photo-novels often make use of such images, and that Coleman can plausibly be seen as taking this over. Thus I observe that ‘one will often find two or more successive moments represented, as if they were occurring simultaneously, side by side within a single frame.’[3] To be fair: I say this explicitly of Coleman’s work; but given that I say it while, as Baetens himself points out, discussing what his work and the photo-novel have in common, any reasonable interpretation of my meaning would have to take my remarks to apply to both. In short: Baetens misrepresents my actual views on this topic. For the record, this is an aspect of Krauss’s account that I still find brilliant: there is no denying Krauss’s gifts as a critic; it is her reliability, as a theorist, in trying to make sense of her own critical insights theoretically that I take issue with.

III. Varieties of Automatism

Palermo’s ‘Automatism’ is a serious response, and there is much in it to agree with. It is, however, highly compact in places and I am not always certain what Palermo is arguing. This makes responding difficult, but I have tried to do so in good faith. Palermo considers not only my own contribution to the special issue, together with a second paper, co-authored with Dawn M. Wilson (née Phillips), but Patrick Maynard’s contribution and book, The Engine of Visualisation, and Dominic McIver Lopes’s Afterword, on which he leans quite heavily in places. Inter alia, he considers the evolution of Peter Henry Emerson’s ‘naturalist’ aesthetics, the history of surrealist automatism and William James’s account of habit. This is a rich mix that goes well beyond his response to my own views, to make an original contribution to the special issue themes of its own.

As regards my own paper, Palermo raises two main criticisms: that my objections to Krauss’s characterization of Kentridge’s working procedures in terms of ‘two kinds of automatism’ are misplaced, when viewed in a broader context than recent theoretical reflection on photography; worse, that they fall foul of my own earlier suggestions, co-authored with Dawn Wilson, about how best to advance theoretical reflection on photography. He also implies, but does not argue, that the conception of automatism I attribute to Cavell is wrongly so attributed. I have something to say about all three in what follows: but as I take the first to be Palermo’s main point, on which the other two devolve, and also the most difficult to bring into focus philosophically, I devote most time to that.

Palermo, I think rightly, takes my criticisms of Krauss’s account of automatism to be largely independent of our deeper disagreements with respect to artistic media. Thus I could have made both the arguments summarized above without mentioning Krauss’s ‘two kinds of automatism’—and perhaps I should have. This is why Palermo can take issue with the latter without taking issue with the former. It is also why, even should he turn out to be right about automatism, my critique of Krauss on artistic media will pass unchallenged. But is he right? Palermo argues that I am wrong to take issue with Krauss’s use of automatism, on the grounds that I do, because Krauss’s use of this idea is in fact ‘perfectly orthodox’—especially when set within a broader historical and cultural context stretching back to the 18^th Century.[4] Seen in this light, beyond the provincialism of recent theoretical debate about photography, the notion of ‘automatism’ has a broad range of meanings that characterizes, among other things: whatever turns stimulus into action, initiates action ex nihilo, or is done instinctively; actions that are performed mechanically or unthinkingly, whether because conventional or because rendered habitual by training; and actions (including mental actions) performed without conscious control, such as free association or automatic writing. So construed, automatism would seem to pick out any action that is not self-consciously thematized as such by the agent, whether because preconscious, unconscious or in principle unavailable to consciousness, but which is something done by the agent nonetheless.

I am not the person to assess Palermo’s historical claims, but I do have some reservations about the wide and disparate variety of grounds on the basis of which such (alleged) actions are supposed to be unthematizable by agents: so generously construed, this threatens to render the idea of an automatism either so pervasive as to be uninformative, or so multivalent as to be self-contradictory. At the most basic level, there is a distinction to be observed between processes that could not be reflectively thematized, because they take place at a motor or neural level that subtends‚ and so is inaccessible to, reflective awareness, and thoughts and activities that are merely no longer or not yet thematized—whether because conventional, habitual or even repressed, though they are in principle capable of being thematized. Call such (putative) actions strongly and weakly automatic respectively.

An example of the former kind on James’s account of habit, to which Palermo appeals for some of his more compelling examples, might be the neural impression created by one muscle’s contraction (allegedly) sufficing to trigger that of the next, in an oft-performed action, or the thinking of one thought being sufficient to prompt its successor, in an oft-rehearsed train of thought.[5] Think of practicing musical scales or setting out a familiar proof. Whatever it is in our physiology that is responsible for the former, or our neurology that is responsible for the latter, they are not themselves subject to the will: we do not intend that one muscle trigger the next, or that one thought trigger another. [6] They simply do, whenever we perform the acts or think the thoughts in question: the latter being something we can, and in many cases do, intend to do. For something like this reason, if I understand him, Palermo wants to call them ‘automatisms.’

But note the difference to examples of the other kind: that some activity is capable of being brought to attention and intentionally corrected shows, by contrast, that it is only unavailable to reflection in a weaker sense. Think of a piano teacher correcting a student’s poor fingering, a tennis coach working on a student’s backhand, or a language teacher pointing up a student’s reliance on a false friend: in each case, doing so brings the fault to reflective attention, making it available for correction. Although a reductive account would no doubt want to explain all three in terms of more basic phenomena—the neural pathways responsible for muscle memory or mental association perhaps—these are not what student or teacher focus on in trying to correct them: they focus instead on how the hands are held or the way in which the keys are depressed; on how the racket-head is rolled, across the full extension of the swing; or on trying to break the habit of reaching for a certain word in a certain circumstances. And it need not be someone else who brings such ‘automatisms,’ if that is what they are, to our attention: a single dead key on a keyboard is enough to bring the otherwise non-reflective activity of typing to thematic awareness, simply by forcing one to remember to strike that key harder.

It is not entirely clear whether Palermo, when he appeals to the OED, means to endorse both the strong and weak senses in which an (alleged) action may be unreflective as sufficient grounds for calling it an automatism—and, if so, whether he regards them as being of a piece, as reducible one to the other, or as a single phenomenon under different descriptions—or simply to enumerate the various, possibly conflicting, meanings that the term has in fact acquired over time. Though it seems to be the former, given that he produces examples of both in order to motivate and, I take it, justify Krauss’s uses of term. By contrast, I see no reason to believe that the stronger—call them ‘structurally unconscious’—senses of automatism constitute a category of action, automatic or otherwise, at least on the account with which he wants to underwrite them. To apply Elisabeth Anscombe’s chief criterion, to whose work on intention Palermo appeals for his positive characterization of an automatism: they could not, being below the level of possible reflective thematization, answer, in the relevant sense of being susceptible to reasons, to the question ‘Why?’ under any description.[7] Because this is the sort of remark that will mean little to anyone not already familiar with Anscombe’s account of intention, or some of the ways in which it has been taken over and adapted by later philosophers of action, such as Donald Davidson, I shall spend some time unpacking it.[8]

Something is an action, on Anscombe’s account, just in case it is intentional under some description, though it need not be under many others, most of which the agent may not even be aware of. To wit: ‘playing a backhand pass,’ not circulating air or swatting flies; ‘playing a Bach Partita,’ not waking the children or bothering the cat; ‘writing my response to Palermo,’ not breaking in a new keypad or straining my wrists. Though the latter may all be things that someone doing the former is also doing, in doing whatever it is that she is doing, she need not be doing so under any of these descriptions—at least not right now, though she might be on some other occasion. If something is not intentional under any description, however, we would be wrong to call it an action on either Anscombe’s or Davidson’s accounts. Imagine receiving, in response to the question ‘Why are you waving your racket around/pounding the keyboard like that?’, the reply: ‘I have no idea,’ ‘I didn’t realize I was,’ or ‘my arms are just flailing about, I can’t seem to stop them’ One would be hard-pressed, in such circumstances, to regard the racket waving or keyboard pounding as an action at all.[9] What I am calling ‘structurally unconscious’ processes or activities, being non-contingently unavailable to thematization, are equally unamenable to the question ‘Why?’ in the relevant sense. For this reason they cannot be actions, automatic or otherwise—at least not on the account at issue.

Given the historically and culturally enriched—but, I have suggested, internally conflicted—conception of an automatism with which he is working, Palermo believes that I am mistaken to criticize Krauss for using this notion to characterize both Kentridge’s habitual working methods and the freeing up of unconscious associations that they make possible. With respect to the former, I am happy to concede that the resources Palermo musters make a compelling case that Krauss’s use of this term is indeed ‘perfectly orthodox.’ Of course, this will hardly constitute a defense if orthodox use is as conflicted I believe. Nonetheless, taken on its own, I find little to disagree with here—assuming that it is also the sense of automatism at issue. In this context, given Krauss’s appeal to Cavell’s authority, this must mean not just whether it is Krauss’s—Palermo has convinced me that it is—but whether it is also Cavell’s; Palermo’s response has made me aware that I could say more on this score. I am also happy to defer to Palermo on the orthodoxy or otherwise, taken on its own, of the latter of Krauss’s ‘two kinds of automatism.’ As I said, I am not the person to query Palermo’s historical claims: indeed, it strikes me as a strength of his paper that it brings out just how common both of Krauss’s uses may be. The pressing question in the context of my original argument, however, would be the extent to which either sense is also Cavell’s. I shall come back to this.

But what is the right conclusion to draw from the fact that both of Krauss’s uses of ‘automatism’ can be supported by past practice? It is here that Palermo and I part company, and perhaps also that our disciplinary differences show up most clearly. Here I think it is worth recalling that the special issue in which my paper appeared was intended to bring philosophers and art historians into dialogue around the point at which otherwise distinct disciplinary debates about photography seem to intersect, namely, with respect to the roles of agency and automatism in the photographic process. I salute Palermo for taking up this challenge so resolutely. Unlike Palermo, however, I do not think that the right implication to draw, from the fact that both of Krauss’s uses of automatism can be independently supported by appeals to past practice, is that Krauss’s use of that term is just fine; I think the implication to draw is that we need to think more carefully about whether one and the same concept can be used in both these ways simultaneously without lapsing into incoherence. My own view is that this is, at best, questionable for reasons that the philosophical framework to which Palermo appeals in trying to make sense of automatism brings out.

To boil things right down, as Krauss uses these terms, in the former sense—call it ‘automatism₁’—automatism essentially means habit; in the latter sense—call it ‘automatism₂’—automatism essentially picks out mental processes (such as free association) that take place independently of conscious control. On Krauss’s account of Kentridge, which Palermo appears to endorse, automatism₁ unleashes automatism₂. But it is not clear whether free association, to the extent that it really is unconsciously determined—rather than merely absent-mindedly performed, but in principle amenable to conscious scrutiny—could count as a kind of act on Anscombe’s account for reasons similar to those already given. Imagine being called to account for producing an (unfortunate) free association mid-conversation. Faced with such a demand, an honest reply might be: ‘I’m sorry, I’ve no idea where that came from; I just heard myself saying it.’ Such a reply amounts to questioning the applicability of the question ‘Why?’ in the relevant sense.[10] So even if Palermo may be onto something in appealing to Anscombe’s account of intention to elucidate Krauss on automatism, he will have a hard time accommodating examples of this kind within such an account. For Anscombe seems—I grant that she is far from clear on this point—to rule out such examples as genuine, or at least straightforward, cases of action; yet automatism is supposed to pick out a particular kind of act on Palermo’s account. Conversely, if Palermo is right to cite examples of this kind as standard cases of automatism, and Anscombe’s account either rules out or raises serious doubts about such cases, then, by parity of reasoning, Palermo must be misguided to appeal to Anscombe’s conception of agency to explain them.

In fact, the issues here are fine-grained and resistant to easy summary. Note, for example, that the standard cases of ‘automatism₂’ are not ruled out on the same grounds as the standard cases of ‘structurally unconscious’ processes. The latter, given that they (normally) take place below the level of what can be reflectively thematized, are (normally) ruled out a priori: one cannot intelligibly ask a person why they are doing something which they can have no awareness of doing.[11] Unconscious associations, by contrast, are ruled out for only so long as they remain unconscious. They are in principle capable of becoming conscious—were they not, there could be no practice of psychoanalysis—and once they do they should in principle be amenable to the question ‘Why?’ in something like Anscombe’s sense. Consider a modified version of my earlier example: ‘I still can’t understand why you would say something like that!’ ‘I didn’t know it/want to acknowledge it at the time, but I think I may have been angry about…’ In such a scenario what originally seemed beyond the scope of reason giving—‘I’ve no idea where that came from; I just heard myself saying it’—is re-described, albeit retrospectively, in such a way as to seem in candidacy for the status of a (mental) action. And that it can be may prove important when it comes to the special case of mobilizing automatisms in the service of an artistic practice.

I suspect that a large part of the difficulty here, at least in coming to a clear view of what Palermo is committed to, derives from the fact that his appeal to the OED generates an implausibly generous conception of automatism that is (at least) tripartite, unlike the bipartite account that it is intended to clarify. Or, to put the same point a different way: to the extent that it is bipartite, it is so on very different grounds to Krauss’s, grounds that would render both of Krauss’s senses subsets of one of his own. For Palermo’s conception of automatism encompasses not only Krauss’s ‘two kinds of automatism,’ but something that Krauss herself does not consider, namely, sub-sentient motor or neural activity. Palermo wants to re-conceive all three, with the help of Anscombe, as species of action. But the latter, what I have called ‘strongly automatic’ processes, cannot—other than in the most exceptional of cases—constitute action as Anscombe would understand this. By contrast, both of Krauss’s senses of automatism, precisely in so far as they are only weakly automatic could—at least in principle—count as actions on Anscombe’s account, other conditions being met. In this respect, Krauss’s own account of automatism is in better shape than the one with which Palermo tries to buttress it. Nonetheless, I shall argue in what follows that automatism₁ is by far the more straightforward case, and that there is good reason to treat automatism₂ with caution, as Anscombe herself does. In sum, Anscombe remains an unlikely ally for making sense of the latter.

This unthematized non-alignment between Palermo’s account and Krauss’s, on the one hand, and his attempt to enlist Anscombe, on the other, renders Palermo’s positive conception of automatism highly illusive, once his examples are subject to scrutiny. Palermo’s first proposal arises in the course of discussing automaticity in general. Automatic machines make the attainment of human goals easier, by automating the labor required to realize them. Take the relation between the desire for clean clothes and washing machines: although washing machines perform their function (once set) without the need for human intervention, far from being impervious to human agency, such machines are modeled on human activity, and designed with human purposes in mind; indeed their very existence presupposes such purposes. Applying such considerations to the photographic case, where the automaticity of the apparatus is often taken to preclude the possibility of fully-fledged photographic agency, Palermo discusses how to understand the act of availing oneself of a camera’s auto-focus function:

If I have a camera that focuses automatically, then, when I press the shutter release, it automatically focuses. That means, it focuses when I press the button. It does not focus accidentally. I expect it to focus, I want it to focus, I intend to take a well focused picture by availing myself of this feature of my auto-focus camera. […] An automatic action is not an accidental one; it is an intentional operation. But, if someone were to ask me what I was doing when I depressed the shutter release, I would say ‘taking a picture,’ not ‘focusing.’ The gist of ‘automaticity’ in this context is nicely captured by contrasting intended actions with descriptions. We call an action automatic if we acknowledge it as our act (as opposed, say, to an unintended consequence) but want to distinguish it from the description under which we perform that action.[12]

An act is automatic on this understanding if it can be distinguished, despite being done on purpose, from the description under which it was performed.[13] But what intended action and description does Palermo have in mind here, and can they be ‘contrasted’ in the way he suggests? I take it that the intentional but automatic act in question is focusing, and the description under which it is performed is ‘taking a [well focused] picture.’ The equivocation between ‘taking a picture’ and ‘taking a well focused picture’ is Palermo’s, and it generates conflicting ways of interpreting what he goes on to say. On one construal, focusing is internal to the description under which such a person acts, and no gap arises between intention and description that needs to be accounted for; on the other, focusing is external to (but compatible with) the description under which such a person acts, creating a gap that automatism is meant to fill. Which construal is correct? Palermo opts for the latter: yet as he describes the case himself, it seems clear that focusing is built in to the description under which such a person acts from the outset.

By his own admission, he ‘intends to take a well-focused picture.’ Asked what he was doing, he reports that he would say ‘taking a picture’ not ‘focusing;’ but is this anything other than an abbreviated way of saying just what he intends? If so, there is no ‘contrast’ between intention and description here, hence no gap for automatism to plug. Despite his intention to keep them apart, Palermo seems to be conflating ‘automatic’—by his own account, a predicate of machines—and ‘automatism’—by his own lights, a kind of act and hence, indirectly, a predicate of agents—here.[14] It is the camera—not Palermo—that focuses automatically.[15] This is hardly news: it is an auto-focus camera! If Palermo himself does anything automatically here, it is not focusing: remaining still, holding the camera steady, being careful not to obscure the lens, or gently applying pressure to the shutter release (so as not to shake the camera), all look like more promising candidates. Asked what he was doing, Palermo would presumably not reply ‘standing still’ or ‘holding the camera steady:’ though both are things one typically does, more or less unreflectively, in ‘taking a picture.’ Such actions still count as intentional under this description because each is done in the service of taking the picture, even though neither provides a description that someone taking a picture would typically use to describe what they were doing.[16] To put the same point in terms that Palermo himself explicitly, but I think mistakenly, rejects: all such acts are ‘swallowed up’ by the description ‘taking a picture.’[17] For this reason, these aspects of taking a picture can credibly be described as both intentional and non-reflective, thereby fulfilling Palermo’s own description of an automatism.

            Indeed, this is precisely the sense of automatism implied by Palermo’s appeal to William James on habit. On James’s account, habit (very generously construed) pervades both simple acts, such as walking, and more complex acts, such as fencing or playing piano, alike. According to James, such abilities lie on a continuum, much of which can be mastered developmentally or instinctively—think of a toddler learning to walk, or a cat learning to hunt—but some of which, as the tasks become more demanding, only through intensive training. In the case of expert performance of various kinds, sufficient drilling can make technical proficiency second nature in the sense that the simple acts which, taken together, go to make up more complex acts no longer need to be consciously selected or reflectively thematized in order to be successfully executed.[18] Think of all the individual finger movements required to play a Bach Partita: despite not being individually thematized by an accomplished pianist, such movements still count as intentional under a description of the more complex act in which the pianist is engaged, because each is done in order to play the whole piece, and that is done intentionally. I am not concerned here with merits (or otherwise) of James’s account, taken on its own terms, but only with the light that Palermo’s appeal to it sheds on his own. What James is describing is in fact a high-level instance of the sorts of automatism—which do not include auto-focusing, but might include focusing or, in the case of auto-focus cameras, registering that the camera has correctly attained focus—that I argued could be involved in ‘taking a [well-focused] photograph.’ Should this sound counter-intuitive, try to envisage a press photographer working at speed in a fast moving and dangerous environment.

Moreover, with respect to Palermo’s example, there is another way to show that focusing is internal to the description under which such a person acts, and thus requires no invocation of automatism to make sense of it. This one also has the advantage of involving no hair-splitting about what such a person might mean by ‘taking a picture,’ and is fully in keeping with the spirit of Anscombe’s account. Anscombe maintains that to find out what a person intends on a given occasion one should look at what they do rather than, per impossibile, aspiring to peer into some private mental event in their heads.[19] This advice can be applied to the present case. Suppose that my camera has been engineered, as is common, to produce a well-focused image of whatever can be seen within a central sub-section of its viewfinder. Assume also that I have a rough and ready sense of this from prior use of the camera. These assumptions being granted, then, simply by doing this—that is, simply by positioning the subject of my photograph centre-stage—while refraining from doing anything that would obviously count against it, such as vigorously shaking the camera, I display my intention to take a well-focused picture in what I do.[20] If this is correct, then talk of a ‘contrast’ between intended action and description is misplaced here. Assuming, in the absence of reasons to think otherwise, that I do not generally want illegible images from my photography—And why would I?—then getting a well-focused image seems to be implicit in the description under which I act.

Perhaps it will be objected that this amounts to agreeing with Palermo: after all, used in this way, ‘implicit’ is a valid substitution term for ‘reflectively unthematized’ (at least weakly construed) and I have suggested that the latter is the one thing that all Palermo’s senses of automatism have in common. But this is not plausible: Palermo not only insists on a ‘contrast’ between intended actions and descriptions, he toughens this up into a ‘disjunction’ or ‘division between action and description’ constitutive of the automatic. One can hardly suppose that all one means by this is that one of the terms is implicit in the other: for what kind of disjunction would that be? This comes out clearly in Palermo’s description of the ‘disassociation’ at the root of automatic writing, his second example:

Planchette writing is a form of automatism not because planchettes are machines, but because the spirit medium is divided (dissociated), so that she cannot describe herself as writing, even if she can acknowledge that she is doing the writing […]

      [T]he automatism lies in [the spirit medium’s] dual relation to her own writing activity. It has nothing to do with mechanism or even with spontaneity, as is generally believed of automatism […] It is the medium’s and the researcher’s shared understanding of automatic writing—according to which it both is and is not the medium’s act—that makes it automatic. Generally, I think we want to speak of an automatic act as automatic if we are tempted to note some disconnection (or disconnections) between our intentions in performing it and the description under which we act.[21]

In the last analysis, perhaps the reason it is so hard to be certain that one has a firm purchase on what Palermo is arguing, has less to do with anything internal to such formulations, and more to do with the fact that they sit so uncomfortably with the philosophical framework with which he wants to underwrite them. Try applying the question ‘Why?,’ in Anscombe’s sense, to the case of automatic writing. The question asks if there is a reason a given bodily movement occurred. If the response provides a reason, by showing that the movement was intentional under some description, then the movement will count as an action rather than, say, a spasm, a piece of compulsive behavior, or an effect to be accounted for in causal terms.[22] The spirit medium’s answer would be that, though it is her hand that is moving, it is not her, but the Spirit, that is moving it. But this amounts to refusing the question application: if, by her own testimony, the spirit medium does not move her own hand, then there is no action of hers to inquire about here. The skeptical researcher’s answer would be that some kind of autohypnosis or ideomotor phenomenon (an involuntary, reflex response to an ideational prompt) is responsible for her hand’s movement. This also amounts to denying the question application, in so far as it states a cause rather than provides a reason.[23]

Consider, more generally, the ‘disconnection,’ ‘disjunction’ or ‘division’ between intended action and description at the root of Palermo’s account in this light. On the theory of action, deriving from Anscombe and Davidson via Lopes that Palermo seems to favor, there just is no such ‘division’ between intentional action and description to be had. On the contrary, being ‘intentional under some description’ is what makes a bodily movement an action, and this entails a strong (indeed necessary) connection between intentions and descriptions, rather than the converse; something is an act only if it is intentional under some description. The same action may be unintentional under many other descriptions, of course, but that only shows that we can do many things that we do not intend to do, and need not even know that we are doing, in doing whatever it is that we do intend to do, and do know that we are doing. If I take a photograph with a camera that has been gathering dust on my shelf for several days, I may unwittingly remove dust from the shutter release button by pressing it. My action is not intentional under the description ‘cleaning the shutter release,’ even though cleaning dust from this button is something that I do in ‘taking the picture’—under which description precisely the same bodily movement is intentional. Could this be the ‘disjunction’ between action and description that Palermo has in mind?

Palermo may be tempted by this prospect, given that cleaning dust from the switch appears not to be ‘swallowed up’ by taking the picture. That is, unlike standing still, or holding the camera steady, we do not take pictures by cleaning dust from switches; just as we do not play backhand passes by circulating air, or Bach Partitas by bothering cats. Put differently, we do not clean dust from the switch in order to take a picture. But this cannot be the right way to go, given that redescribing an act in such a way that its agent no longer recognizes it as a description of something that they set out to do, even though they grant that it is something that they did, is just what makes an act unintentional on Anscombe’s account; and Palermo maintains that automatic acts are intentional, despite being distinguished from the descriptions under which they are performed.

Perhaps Palermo is after a more restrictive claim: one that distinguishes automatic actions from ‘straightforwardly’ intentional and unintentional actions alike? Here something basic to both Anscombe and Davidson’s theories bears emphasizing that Palermo appears insufficiently sensitive to the consequences of: actions as such—that is, in themselves—are neither intentional nor unintentional. As a corollary, there can be no class of intentional or unintentional acts: rather, the same act may be intentional under some descriptions and unintentional under many others.[24] Take Davidson’s classic example: I move my finger, flipping a switch, turning on the light, illuminating a room. Without meaning to, I thereby alert the prowler who happens to be lurking in the room.[25] Whether I intended to do so or not is irrelevant: alerting the prowler is something that I did, by turning on the light. What might talk of a ‘disconnect’ between intention and description pick out here? There is none between my flipping the switch and illuminating the room. Asked what I take myself to be doing, I might reply: ‘turning on the light.’ My action is clearly intentional under this description. It is not intentional under the description ‘alerting the prowler’ of course; but it would be a very odd way of speaking to say that I alerted the prowler automatically—as strained as saying I cleaned dust from the shutter button automatically—especially given Palermo’s own gloss on the meanings of the latter.

To wit: alerting the prowler is neither habitual nor conventional; it is not an act performed unreflectively, like reaching for the door handle when leaving a familiar room, or out of deference to social convention, like uttering certain formulaic greetings on running into someone one knows. Neither is it instinctual, like a newborn rooting for the breast, nor something one might be so intensively drilled in doing as to have become instinctual—like parrying an offensive lunge in fencing. Nor is alerting prowlers the sort of thing it makes sense to think one may be so well-trained in doing as to no longer need thematize to oneself all the intermediate acts required to carry it off—like playing a particular piano piece. It is also not something one does ‘unconsciously,’ if we wish to avoid equivocating on standard meanings of that term. Granted, I don’t do it wittingly. How could I: I didn’t know that the prowler was there! But doing something unknowingly is not the same as doing it unconsciously. The person who acts unconsciously does not do so for lack of information about the world on a particular occasion: what they typically lack is a form of self-knowledge that has implications for their life (and hence all their acts) as a whole. Indeed, this very diversity of phenomena that ‘automatism’ is asked to cover on Palermo’s account should give us pause.

But perhaps all this is a distraction; perhaps all these possibilities were supposed to have been ruled out in advance by Palermo’s qualification that ‘[w]e call an action automatic if we acknowledge it as our act (as opposed, say, to an unintended consequence) but want to distinguish it from the description under which we perform that action.’[26] Prima facie, alerting the prowler would seem to be as good a candidate as any for an ‘unintended consequence.’ Set aside, for the sake of argument, the Feinberg-Davidson notion of an ‘Accordian Effect,’ according to which, whether itself intended or not, anything that is a consequence of something that I intend under some description (as I here intend to illuminate the room) is something for which I may be held responsible as an agent.[27] Even setting this to one side, however, the problem is that under these circumstances, flipping the switch just is to alert the prowler: It is not a consequence of my act in the way that, were the sudden illumination to startle the prowler, causing him to trip over a coffee table and knock himself unconscious, the prowler knocking himself out would be an unintended consequence—for which, if Davidson and Feinberg are right, I may be held responsible as agent. On the contrary, alerting the prowler is a redescription of my turning on the light such that, though I can still recognize it as something that I did, I no longer recognize (under the new description) the intentions with which I did it. By contrast, my act of turning on the light cannot be redescribed as knocking the prowler unconscious, though it did cause it. For this reason, alerting the prowler cannot itself be ruled out as an unintended consequence: it is not a consequence, intended or otherwise. If the case is ruled out on Palermo’s conception of an automatism, it can only be in virtue of being unintentional. For just the same reason, however, most standard cases of automatism₂ would also appear to be ruled out.

If Davidson’s example leaves room for a ‘disconnect’ between intended action and description at all, it can only reside in what gets done unreflectively in carrying out whatever it is that we intend to do. In this case: taking up a position a comfortable distance from the wall, lifting my right arm while extending my right index finger, and applying pressure to the switch. All are acts I typically perform in turning on the light, yet none furnishes the description under which I typically act in doing so. Asked what I was doing, I would not reply: ‘raising my arm,’ ‘extending my index finger’ or ‘applying pressure to the switch,’ though all three would be true. If anything does, these fulfill Palermo’s description of an automatism as something we are willing to ‘acknowledge as our act’ despite ‘distinguish[ing] it from the description under which we perform that action.’ If anything is, these seem to be (low-level) examples of the sort of thing James has in mind when he theorizes habit as the replacement of acts that were once carried out with laborious deliberation with automatisms that have since become second nature through repetition. Try asking a toddler to turn on a light.

On the face of it, this ought to be a happy result. Here is a phenomenon that unequivocally fulfills all Palermo’s positive remarks about automatism₁. The problem is that Palermo himself explicitly rules out this way of understanding an automatism, despite the fact that it is the natural way to understand much of what he says, and the fact that his own cases of automatisms taking the form of high-level, well-drilled performance would seem to require it. Palermo rules it out in a rather coy footnote, appended to the first of my two citations above, in which he refers the reader to what can be ‘glimpsed’ in §26 of Anscombe’s Intention. Here Anscombe considers actions that are not ‘swallowed up’ by the final description in which they figure. This requires some unpacking, not least because it is the only point at which Palermo invokes Anscombe directly, though, as the citations above attest, her terminology permeates much of what he says.

In the example on which Anscombe is commenting, by moving his arm up and down while gripping the handle of a working pump, a man pumps water, replenishing the water supply of a house, and (it being poisoned water) gradually poisons the house’s inhabitants. Palermo’s suggestion is that what makes an act automatic is that it is not ‘swallowed up’ by richer descriptions of the same act in the way that thinner descriptions of that act are. Consider the implications of this claim. Tweaking Anscombe’s example, imagine the following exchange between co-conspirators: ‘Why are you moving your arm like that?’ ‘Can’t you see, I’m pumping!’ ‘Of course I can see that: why are you pumping?’ ‘Why do you think? I’m replenishing the water supply.’ ‘Why bother? You know we’re going to kill them all before we leave anyway.’ ‘But I am killing them!’ Anscombe’s thought would be that the narrower descriptions of what the active man is doing (moving his arm, operating the pump, replenishing the supply) are ‘swallowed up’ by the broadest description of what he is doing; poisoning the inhabitants. The same is true on Davidson’s account: the Queen turns her wrist, empties a vial, tipping poison into the ear of the King, killing the King. The first three are ‘swallowed up’ by the last, killing the King.[28]

Note that on both Anscombe and Davidson’s accounts there is in fact only one action here—that of moving one’s body in a certain way—and that one action fulfills the later, more complex descriptions in the series in virtue of being considered within an ever wider net of circumstances within which it takes place. In these circumstances—holding the handle of a working pump connected to a poisoned water source supplying a house; holding an open vial of poison over the King’s exposed ear—the acts of moving one’s arm up and down or rotating one’s wrist just are, respectively, poisoning the inhabitants or killing the King. The broadest description in each series picks out the intentions with which the act in its narrower descriptions was undertaken, thereby ‘swallowing them up.’ That it does is shown by the fact that we can give the answer ‘poisoning the inhabitants’ or ‘killing the King’ to the question ‘Why are you moving your arm/wrist like that?’ Similarly, ‘taking a picture’ picks out the intentions with which one stands still, holds the camera steady, gently applies pressure to the shutter release and so on. As such, they are swallowed up by that description. Conversely, the narrower descriptions tell us how the act under its broadest description was carried out—by standing still, holding the camera steady, gently releasing the shutter, and so on.

By contrast, were the man to beat out the rhythm of ‘God Save the Queen’ on the pump by modulating his pumping action in such a way that the pump’s habitual clicking noise beats out that tune, this would not be swallowed up by the forgoing series. We cannot give the answer ‘poisoning the inhabitants’ to the question ‘Why are you beating out the rhythm of God Saves the Queen on the pump?,’ since we do not poison the inhabitants by beating out that rhythm. A more plausible answer, Anscombe notes, might be: ‘I discovered I could do it, and since I have to pump anyway, I do it to relieve the tedium.’ So were one to receive the reply, ‘To beat out God Save the Queen,’ to the question ‘Why are you moving your arm up and down like that,’ asking ‘Why?’ of that reply in turn would not lead to ‘poisoning the inhabitants,’ and so would not be ‘swallowed up’ by that description. But it would seem very strained to say that the man beats out the tune automatically, as we should if Palermo is right.

To be banal: it is hard to see how this could count as an ‘automatism’ if, as seems likely, the man has to focus his attention on doing so, so as not to make a mistake. Should this objection be met with the response that he may now be so practiced at doing so that he can do it while thinking of something else entirely—perhaps how he intends to spend his ill-gotten gains—what this reply reveals is that the act is being conceived as automatic on more straightforward grounds of habit. At that point, not being ‘swallowed up’ by the broadest description of what he is doing falls away as doing no explanatory work. Conversely, should it meet with the reply that the man is beating out the tune unconsciously or unwittingly, and assuming enough more is said to allow us to make sense of such a response in this case, I have already shown why the case would be ruled out on an account grounded, like Palermo’s, in Anscombe. If an automatism is an intentional act, it should be susceptible to the question ‘Why?;’ but we cannot intelligibly ask why someone is doing something of which they have no awareness.

With respect to Palermo’s main objection, then, I conclude that a single term cannot be used simultaneously in all the ways that Palermo’s appeal to the historical record shows that ‘automatism’ has in fact been used without lapsing into incoherence. I have already shown that strongly automatic processes cannot count as actions in Anscombe’s sense. But neither Palermo nor Krauss can have this both ways. Restricting one’s focus now to the weakly automatic: if one sense of automatism is, as Krauss claims, ‘the very opposite of the first,’ what this should tell us is that there are two conflicting concepts at stake here, not one.[29] Given his appeal to Anscombe, I suggest that Palermo would do best to give up automatism₂. An automatism is supposed to be a peculiar kind of intentional act on Palermo’s account, but it is hard to see how automatism₂ can standardly count as such, on the account with which he tries to underwrite it, since it cannot be made to answer to the question ‘Why?’ without special pleading. By contrast, Anscombe’s account provides a perfectly straightforward way of understanding automatism₁. If, as I suspect, Palermo is unwilling to give up automatism_2, given the orthodox nature of this usage in art history, then he ought, in all consistency, to give up trying to account for it in terms of the tradition of action theory that Anscombe’s idea of an act being intentional or otherwise only under some description has given rise to; since this tradition would seem to rule out this use. Failing this, Palermo’s account will remain as conceptually conflicted as the historical discourse it models.

Should Palermo still want to use Anscombe, I suspect that the resources that may serve his purposes better reside elsewhere in her account entirely, in the distinction between an act being intentional and merely voluntary. The latter is a weaker requirement: all intentional acts are voluntary, but not all voluntary acts are intentional. In addition, they must be amenable to the question ‘Why?’ in the reason-giving sense. Automatism₂, as we have seen, does not standardly rise to this more stringent demand. Could automatism₂ pick out what Anscombe means by the ‘merely voluntary’ as opposed to the intentional? Though it remains one of the murkiest aspects of her account, and not something I propose to comment on further here, it at least bears noting in this context that Anscombe isolates ‘merely voluntary’ acts by drawing attention to just those cases in which we are inclined to say things like: ‘I don’t know why I did it’ ‘I just found myself doing it.’ ‘I heard myself say…’ and so on. Of such cases, she remarks: ‘the question “Why?” has and yet has not application; it has application in the sense that it is admitted as an appropriate question; it lacks it in the sense that the answer is that there is no answer.’[30] Such cases seem a much better fit with what most authors, Palermo included, seem to have in mind by automatism₂. Being the terrain that psychoanalysis classically studies—slips of the tongue, rationally inexplicable acts or omissions—this looks like a much more promising basis for automatism₂. But it is important to grasp the exceptional status of such cases for Anscombe, given the ways in which they depart from her basic account of being intentional only under a description; and it is on the latter that Palermo relies. Thus, even if merely voluntary acts are conceived as borderline cases of action, given that they are not generally amenable to the question ‘Why?,’ they cannot generally count as intentional: yet automatism is supposed to pick out a kind of intentional act on Palermo’s account.

IV. Agency and/or Automatism

I shall respond to Palermo’s two other criticisms more briefly, as much of what I need to do so is now in place. In concluding, Palermo implies that my objections to Krauss’s understanding of Kentridge’s working method in terms of ‘two kinds of automatism’ imply a zero-sum conception of the relation between agency and automatism—such that if an apparatus or process is fully automatic, it must be nully agential, and vice versa—that falls foul of my earlier recommendations for advancing theoretical reflection on photography. In an earlier paper, Wilson (née Phillips) and I contend that it is precisely this underlying tendency to understand the relation between intentionality and causality, like that between other oppositions—including agency and automatism—that structure philosophical reflection on photography, in zero-sum terms that remains one of the greatest barriers to a better understanding of photography as an artistic medium, technical apparatus and optico-chemical process.[31]

Palermo offers little by way of argument to establish that my objections to Krauss do fall foul of my earlier claims about advancing theoretical reflection on photography, satisfying himself with a juxtaposition of citations, but I think his underlying rationale for believing this is clear enough. In my paper on Krauss I contrast Kentridge’s agency as a draughtsman with the automatic nature of the stop frame movie camera. More specifically, I claim that Kentridge’s practice is founded on a ‘structuring tension’ between ‘the ebb and flow of his line,’ which I gloss as ‘an autographic mark in the autographic art par excellence’ that records the hesitations, reversals and discoveries of Kentridge’s draughtsmanship, on the one hand, and the automatic recording of that mark by the ‘impassive’—or what Cavell might call ‘inhuman,’ ‘sterile’—eye of the stop-frame animation camera, on the other.[32] I stand behind this claim: indeed, it seems to me that it is Palermo who is falling foul of his own (very helpful) distinction between automatic and automatism here.

The stop-frame camera that records Kentridge’s incremental additions to his drawings is automatic in the relevant sense: that is what makes it a machine for recording the visual field. Kentridge knows, so long as he does not clatter the camera, that once he has secured it on a tripod, focused on the drawing surface in such a way as to fill the picture’s frame, and set the camera’s exposure correctly, that he need not look through the viewfinder again—at least not for the purposes of getting a correctly exposed, well-focused and well-framed image. Assuming the camera is working properly, this is something that he can rely on. The camera will not reset itself, and it has no mind of its own to change. Kentridge may look through the viewfinder as often as he pleases to see how the drawing may look in the finished work, and thereby take instruction for working on it further, but that is a separate matter. The issue at stake is that no further draughtsman is required to record—or, strictly, testify to—Kentridge’s modifications in turn, and so on ad infinitum; the camera does all this automatically, thereby arresting any potential regress. If it did not, Kentridge’s art form, ‘drawings for projection,’ would not be a live possibility.

On my understanding, the camera is a machine rather than an agent, and Kentridge is an agent rather than a machine. Being an agent and being a machine are antithetical because agents act on reasons whereas machines simply do what they are engineered or programmed to do. One cannot intelligibly ask why, in the relevant sense, a machine does what it does: it simply does what it is designed to do. But this will only be taken to imply that agency and automatism are in a zero-sum relation if one has already conflated automatism with the automatic, but that is something that Palermo himself cautions against doing. Indeed, it is one of the merits of his paper that it insists on a much sharper distinction between these terms than is common. ‘Automatism,’ as Palermo asks us to understand it, picks out something done by agents, rather than a characteristic feature of machines for recording light arrays. In the context of recent debates about photography, in which ‘automatism’ is typically taken to be synonymous with ‘automatic,’ and predicated more or less interchangeably with ‘mechanical’ of the photographic apparatus, distinguishing between the two in this way is striking. I have argued elsewhere that, until quite recently, the default understanding of these issues in philosophy generated a misleading picture of photography in general.[33] I won’t recapitulate my reasons for doing so here; about this, I take it, Palermo and I agree.

But if one believes that automatic and automatism are neither the same, nor even mutually implicating, then claiming that being an agent is incompatible with being automatic—in the sense in which a machine may be automatic—has no implications for the relation between agency and automatism; it certainly does not entail that agency is at odds with automatism. There may be many aspects of Kentridge’s drawing process that are automatisms in one or other of Krauss two senses.[34] This is not something I deny in my original article. But both will belong to the ‘autographic’ dimension of Kentridge’s art, as I understand this.[35] That is, both will be internal to what he does as a draughtsman, hence part and parcel of what makes a Kentridge a ‘Kentridge:’ namely, that it was drawn by him and not by somebody else. That a work’s causal history matters in this way is just what makes it autographic. So I am not denying what Palermo takes me to be denying; I am distinguishing Kentridge’s activity as a draughtsman—including the ways in which his working procedures may be designed to outwit himself by circumventing conscious control—from the automatic nature of the camera as a recording apparatus, which Kentridge must rely on to capture the modifications he makes to his drawings over time. The latter is something that Palermo tends to lose sight of. Take these in turn.

Who knows how Kentridge comes up with just those images and sequences that he does? Whatever the explanation—and irrespective of whether it appeals to automatisms of either kind—all this is part of what Kentridge does qua draughtsman: in so far as it is done in the service of making a drawing, and this is something that Kentridge does intentionally, it counts as an expression of Kentridge’s artistic agency. At this point it will be rightly pointed out that I have maintained throughout that automatism₂ is (at best) a borderline case of agency, and then only if certain other conditions are met. If this is correct, how can I now maintain that, in certain circumstances, automatism₂ does count as an expression of (artistic) agency? To see this, one must distinguish between the levels of fine-grain at which a given act is described. Practices such as Kentridge’s show that automatism₂ can, other conditions being met, contribute to a higher order action. That is, an action described at a higher level of generality. If Kentridge sets out to harness his own automatisms in the service of drawing, if he intentionally sets things up in such a way as not only to facilitate, but actively to solicit and recruit his own automatisms, then the fact that he elicits or stokes his own automatisms will count as an action under the broader description ‘making a drawing.’ Were this not so, Kentridge could not be understood as making art. Kentridge may be, and probably is, acting under some such description as ‘make a drawing that you cannot fully anticipated in advance.’ But this does not entail that the individual automatisms elicited, taken in isolation, count as actions in their own right—even if Kentridge’s act of eliciting them, the fact that he sets them in train, does count as an action, indeed one that is essential to realizing the drawing.[36]

Cameras, by contrast, are (more or less) automatic machines: they need not practice or strive to make the recording of light arrays second nature; this is simply what they are engineered to do. There is nothing to be gained by denying this: it is an excessive reaction to an equally excessive denial of the possibility of photographic agency that starts from recognizing this fact about cameras. Rather than denying the fact, the challenge is to block the skeptical conclusion about the possibility of fully-fledged photographic art that it has often be taken to warrant. For this reason I am happy to endorse the conclusion that Palermo presents as if it were an argument against me:

[N]othing stands as a better example of “remaining open to, indeed setting things up in such a way as to solicit and encourage” an automatic outcome than the agency of the photographer. If Kentridge’s work fails to count as automatic, and fails precisely because he does too much to enable its automaticity, then nothing is less capable of being automatic than photography.’[37]

Nothing I say in my original article should be understood as contesting this, and much of what I say supports it. The only quibble I have is that I would want to distinguish, following Palermo’s own lead, more carefully than he does himself here between automatism and the automatic. In the remark cited above, Palermo is taking issue with me for claiming that Krauss mischaracterizes the nature of Kentridge’s agency in setting things up in such a way as to harness his own free associations as a resource for making art. But this should not be construed, as it is by Palermo, as a conception of artistic agency opposed to automatism. On the contrary, I grant the force of Krauss’s interpretation that Kentridge’s practice is founded upon mobilizing his own automatisms. Likewise, I have no objection to the idea that photographers employ cameras, more or less automatic machines for recording light arrays, as tools for realizing their artistic ends.[38] But these should be recognized as the distinct claims that they are: the former concerns the nature of Kentridge’s practice, specifically what he does and does not do reflectively in realizing his art, given his reliance on his own automatisms; the latter concerns the employment of an automated, labour saving device for producing an image of the world. The conclusion that Palermo presents as if it were an argument against me runs the two together, thereby flouting his own helpful distinction between an automatism as non-reflective action and the automatic as automation.

Suppose I have shown that contrasting agency and the automatic does not—in itself—entail contrasting agency and automatism: though, of course, I have also done this, on independent grounds, for some stretches of what Palermo means by automatism, notably automatism₂, here. Does the strong distinction I draw between agents and machines commit me to a zero-sum conception of the relation between agency and the automatic? No, for reasons that Anscombe and Davidson make clear, and which it is Dom Lopes’s great service to have shown the relevance of for these debates. Lopes’s target, in his afterword to the special issue, is what he calls the ‘material conditional’ governing the inferential patterns of much recent theorizing about photography. The conditional, as Lopes presents it, is a mutually inflected triad comprising medium, automaticity and agency and runs: ‘if what specifies photography as an art medium is its automaticity, then taking advantage of that automaticity curbs authorial agency.’[39] The conditional can be run in either direction, hence, conversely: if what specifies photography as an art medium is its automaticity, then exercising one’s agency as a photographer compromises the photographic nature of one’s work. Lopes’s material conditional is a perfect embodiment of the kind of zero-sum thinking that Wilson and I, in ‘Automatism, Causality and Realism,’ claimed was hindering progress in the philosophy of photography, namely: if an apparatus (such as a camera) is fully automatic, then it must be nully agential; if a process (such as chemistry or optics) is entirely causal it must exclude intentionality—and vice versa. Lopes’s triadic version has the advantage of making the implications of these oppositions for thinking about photography as art explicit. According to Lopes’s version, the more photographic art cleaves to what makes it photography, the less agency can be involved, and vice versa.[40] Strictly speaking, ‘photographic agency’ must be a misnomer: photographic agency would be a peculiar kind of passivity such that one is most authentic, photographically, when one stands back and lets the process and the apparatus work their magic, without interfering. This is a curious state of affairs; in what other domain would something like this be thought true: Spirit mediumship perhaps?

Lopes brings out what is wrong with this view by showing that whether or not anything, including taking a photograph, counts as an action, and hence as an expression of agency, can vary despite holding the automaticity of the camera constant; that it can demonstrates that agency and the automatic cannot be a zero sum relation.[41] Compare the following three cases: I set my camera on the table and press the shutter (and the rest follows); reaching distractedly for my coffee cup I flub, inadvertently hitting the shutter release of the camera next to my cup (and the rest follows); just as I am sitting down, my son takes a flying leap into me, causing me to fall on the camera, tripping the shutter (and the rest follows). Suppose all three result in identical images. The former is clearly my act; I took the picture on purpose. The latter is clearly not: falling onto a button, because crashed into is not something that I do but something that befalls me. The middle case is more ambiguous: it is an accident, and in the case of accidents—at least on Davidson’s analysis—I do one thing unintentionally, while trying to do something else intentionally.[42] So it is my act: I did indeed press, rather than fall on, the button; but I did not do so intentionally. That is, I applied pressure to the shutter release under the description ‘reaching for my coffee cup’ and not under the description, ‘taking a picture.’ The upshot is that holding a strong distinction between agents and machines—as implied by my original claim that Kentridge’s practice pivots upon the tension between his autographic mark and the impassive eye of the camera—commits me to zero-sum conceptions of neither agency and automatism nor agency and the automatic. 

V. Automatism and the Automatic

A final word, in closing, about Cavell: my original article asked whether Krauss can get what she needs from Cavell. In passing I claimed that, despite her appeal to Cavell’s authority, Krauss’s use of ‘automatism’ does not speak to the chief sense of this term in Cavell’s work on photography, namely, the automatic nature of the camera itself. It never occurred to me that this might be considered contentious enough to require an argument. Palermo’s reply suggests otherwise. To be precise, I say that it ‘points to how different Krauss’s use of the language of automatism is from Cavell’s’ that ‘the brute automatism of the camera itself, the fact that cameras are capable of producing an image of whatever they record without subjective mediation—which would be the most obvious sense of automatic for Cavell in this domain—does not even figure.’[43] Palermo thinks that it is misleading to speak of automatism ‘as inhering in or being a feature “of the camera itself”’ for the kind of reasons already discussed, and I am happy to defer to him on this, having learnt from his reply. [44]  But it should be obvious that what I meant here is what Palermo would call the automaticity of the camera and not, to coin a phrase, agent automatism. Palermo characterises the view I attribute to Cavell as ‘a more or less unanalyzed innovation of a certain kind of theorizing about photography’ which he ‘do[es] not see much reason to ascribe to Cavell, or at least, to understand his remarks about automatism in such terms.’ [45]

But this smacks of moving the goalposts: The first half of this remark effectively acknowledges that I mean automatic; the second half implies that I mean automatism. It is true that I muddy the waters myself here, by using automatism and automatic interchangeably. I have already granted the force of Palermo’s considerations against doing so: indeed, it is only in light of reading Palermo that I have been motivated to distinguish between the two in this way. But it should nonetheless be clear that I am talking about the camera and not an agent here; and even a passing familiarity with Cavell’s ontology of film reveals it to be erected on a conception of the relation between a photograph and what it is of (namely, ‘reality’) for which the automatic—non-subjective, non-agential—nature of image production in photography is fundamental. Palermo must know this.

Indeed, Cavell’s conception of the relation between the automatic and automatism is, in spirit if not in letter, in this respect not unlike Palermo’s own. When Cavell discusses the automatic in relation to photography in The World Viewed it is primarily the ‘automatic’ or ‘mechanical’ nature of the camera as a device for producing images of the world that he has in mind. Conversely, when he talks about automatisms (or ‘media within a medium’) in Must We Mean What We Say? and The World Viewed, it is the conventions that an artist working in a given genre or medium must master, in mastering her art, and the question of whether there are any such standing conventions for arts in their modernist guise that he has in mind. These are distinct topics: film is a ‘succession of automatic world projections,’ because rooted in the automatic transcription of appearances by the camera; artistic conventions are automatisms, because something which practitioners of a given art form must master in mastering their art. The former is a predicate of a machine, in this case a camera, the latter is a predicate of an agent, in this case an artist. In this respect at least, the parallels with Palermo’s way of distinguishing between automatism and the automatic are striking:

[T]he physical mechanisms of camera and projector […] produce the physical or material basis of the medium of film, which I am articulating as successions of automatic world projections. What gives significance to features of this physical basis are artistic discoveries of form and genre and type and technique, which I have begun calling automatisms.[46]

Responding to his critics in ‘More of the World Viewed,’ Cavell makes clear that in using the term ‘camera’ to stand in for the totality of these mechanisms he had in mind a generous conception of ‘the entire physical apparatus which comes between what is before the camera and what results after it, on the screen.’ Of the apparatus so construed, he remarks that it is: ‘sterile,’ ‘inhuman’ and ‘automatic;’ ‘has no choice either over what is revealed to it or what it reveals;’ and ‘does all and only what it is made to do.’ In sum, that it is a machine with ‘no conception whatsoever of its own.’[47] Conversely, the ways in which these ‘possibilities of the physical medium of film’ are mobilized both by different film-makers and in different genres of film constitute the medium’s ‘automatisms,’ which Cavell glosses in turn as ‘bearers of the film maker’s intentions.’[48] So understood, film’s automatisms pick out the various ways in which the possibilities afforded by its automatic substrate can be articulated in such a way as to imbue their exploration with artistic significance, and thereby transform an otherwise inert set of physical processes into a medium for art.

At this point it may be observed that while these two topics, automatism and the automatic, are importantly distinct for Cavell, it is equally important to see how they are related. Moreover, doing so explains why Cavell is willing to court confusion by using cognate terms to talk about phenomena at distinct levels of analysis, the material substrate of film, on the one hand, and the conventions through which that substrate is rendered artistically significant on the other. And this is true. Directly after distinguishing between the automatic and automatism in the remarks cited above Cavell says as much himself, noting that it arises as much from the nature of film, specifically, that ‘its medium just does have this manufacturing mechanism at its basis,’ as from the more general conditions of modernism in art, notably ‘that its awareness and responsibility for the physical basis of its art compel it at once to assert and deny the control of its art by that basis.’[49] But while the observation is a useful corrective, and explains why the difficulty arises specifically in the case of film, nothing Cavell says in justifying his procedure detracts from the basic point. Granted, there may be good reason for Cavell to use related terms, but those terms point, on the one hand, to a set of physical processes and procedures, and on the other, to the articulation of those processes and procedures in various characteristic and not so characteristic ways. So the distinction stands.[50]

In emphasizing the automatic nature of the photographic apparatus, or what he calls the ‘inescapable fact of mechanism or automatism in the making of [photographic] images,’ Cavell is responding to André Bazin’s stress on the etiology—as opposed to appearance—of photographs. On Bazin’s account, it is because cameras automatically record their subjects’ reflected light, unsullied by the infelicities and doubts that subjectively rendered images introduce—and, presumably, because we know this—that we invest photographs as a class of images with ‘being of the model’ and in so doing treat them as both ‘closer’ to what they represent, and a more objective record of it—however poor a given photograph may be in informational terms. As Bazin puts it: ‘For the first time, between the originating object and its reproduction, there intervenes only the instrumentality of a nonliving agent. For the first time an image of the world is formed automatically, without the creative intervention of man.’[51] Cavell takes over this stress on the automatic, non-subjectively mediated relation between a photograph and what it is of, because it allows him to take photography (and, by extension, film) as grist to the mill of his motivating concern with philosophical skepticism. As Cavell understands the relation between the two, photography realizes the skeptical wish to transcend the limits (construed as limitations) of human finitude, so as to experience the world, as if from nowhere, free from the constraints of human subjectivity. What photography achieves is unprecedented in the history of representation: it ‘over[comes] subjectivity in a way undreamed of by painting […] by automatism, by removing the human agent from the task of reproduction.’[52]

Cavell’s conception of the relation between photography and skepticism falls beyond the scope of this paper. Nonetheless, the important point for present purposes should be clear: photography achieves this revolution by entrusting the reproduction of appearances to a machine, however broadly construed. That is, a non-agential explanation of the genesis of photographic images is necessary to getting Cavell’s understanding of film as a ‘moving image of skepticism’ off the ground.[53] Absent such a conception, it is impossible to grasp what Cavell wants from photography—and it is clear that he wants something. For this reason I propose that Palermo’s third criticism, the implication that my objections to Krauss are based on a misreading of Cavell, is also unfounded.[54]