Comments on Isabelle Graw’s Concept of Painting

Our concept of painting is notoriously vague. Arguably, any of our categories and concepts of art are. But the theoretical means used to characterize and essentialize painting over the previous several decades is all too well known; today it’s uncontroversially understood1 that painting constitutes almost anything, if it does not already constitute everything.

The current state of affairs decouples painting from its medium-specific properties—painting, rather surprisingly, does not possess any. No longer philosophically medium-specific, nor ideologically concerned with its own medium-specificity, painting can be construed as a medium-unspecific art, polysemous, and beside itself.2

We might, however, put pressure on this orthodoxy by asking what a medium-unspecific notion of painting actually looks like. How does one begin to construct an understanding of painting under such conditions beyond theories of resemblance? How might an art form, supposedly resistant to any definitional properties, be given definitional treatment? Certainly, painting possesses some unique medium-specific properties. If not, it would be utterly unrecognizable, its identification an impossible task. And obviously, this is far from the case.

Yet, we need not be reactionary Greenbergians, nor philosophic dogmatists, to concede that not every artwork is a painting; that some artworks constitute paintings, and others, not. For once this is granted, then we may also, importantly, distinguish painting from other categories of art. Acknowledging this simple banality allows us to see that we do have certain grounds—certain definitional stipulations and reservations—we uphold when it comes to painting (mutatis mutandis our other categories of art). A medium-unspecific definition of painting will thus have to be plastic enough to survive the post-medium principles of Rosalind E. Krauss, yet rigid enough to evoke the seemingly obsolete essence of painting qua painting.

Art historian and critic Isabelle Graw offers up just such a definition, addressing this seemingly contradictory task3 head on, proposing this definition: “Painting is a form of production of signs that is experienced as a highly personalized semiotic activity.”4

While Graw’s definition is by no means universally shared by art theorists, critics, or even artists—capital-“P” painters and non-painters, respectively—it is certainly enjoyed by them, and more often than not, our descriptions and ordinary talk of painting echo much of what is captured in Graw’s theory. But Graw’s definition is interesting for other reasons as well. It is, as far as I know, the only serious definition of painting to be advanced after the Kraussian-shift in our art-theoretic environment. However, as I show in this text, Graw’s definition, while commendable, is flawed—irresolvably so.

My goals for this text are basic. First, I want to briefly explore the elasticity of Graw’s painting concept. I’m interested in how Graw’s definition holds up, not only intuitively, but modally as well, i.e., how it handles certain counterfactual scenarios. Once this is done, I want to show systematically how and why Graw’s definition breaks apart and why it ultimately fails on both intuitive and logical grounds. Less explicitly, I hope to very preliminarily endear the skeptical reader to an idea of painting that is medium-specific without modernist jurisdiction.


1. Medium-unspecific

In Graw’s eyes, painting, today, is best understood as a production of highly personalized Peircean-like signs.5

While Charles S. Peirce’s concept of indexicality is typical to photographic and filmic discourses, it also, according to Graw, extends to painting inasmuch as signs may be taken as an index of traces of the person (of the painter who produced the painting), and this specific process is somehow unique only to painting.

By “highly personalized” Graw means to compare paintings to people, that is, to treat paintings as “quasi-people.”6 Acquiring a painting essentially means that one is, in some way or another, acquiring a part of that person: “Buying artworks indeed comes close to buying people—and this is especially true for painting” (2012, p. 47).7 Not only this, but painting’s longevity, its uncanny ability to stick around, to zombify, is—Graw thinks—largely due to this special quasi-person status. Any definition for painting should thus, per Graw, incorporate this status.

For the moment, let’s take “highly personalized” to roughly mean a “psychological event taking place in an agent wherein which the presence of a living being is evoked qua artifact- or art object-kind.”8

Could we then not easily think up counter-examples, that is “impersonal” methods of artistic production (say, these methods of production are mechanical) that challenge Graw’s definition? Methods that are indirect and removed, where no human being, body, hand, torso, etc., is involved in the creation of the work?

Graw argues that even impersonal methods of production like these are still, inevitably, highly personalized. Gerhard Richter’s squeegee-produced paintings, for example, are still a record of his presence for the reason that the squeegee apparatuses involved indirectly capture his body movements (used to make the painting). In fact, the more absent, or non-existent, an artist is from their work, the more their “signature” or “presence” is reinforced.9 Graw explains:

For this indexical effect to occur, the artist does not need to have literally set her hand on the picture, or to have brandished a brush, or to have thrown paint on it. A mechanically produced silkscreen by Andy Warhol, who often delegated his work to his assistants, or a printed black painting by Wade Guyton, is no less capable of conveying the sense of a latent presence of the artist—by virtue, for instance, of imperfections deliberately left uncorrected, selected combinations of colors, or subsequent improvements. Painting, then, would have to be understood as the art form that is particularly favorable to the belief—widespread in the visual arts more generally—that by approaching or purchasing a work of art, it is possible to get a more immediate access to what is assumed to be the person of the artist and her life. (2012, p. 51-52, my italics)

But, we may remain unconvinced. How is it possible for a perfectly executed, mechanically-produced painting, that admits no technical imperfections, to plausibly constitute a highly personalized semiotic activity?10 For example, a painting that is produced entirely mechanically, down to the choice of color, composition, etc., will seem to convey no trace of the artist, her “presence,” and so on, and will have avoided the indexical effect Graw has prescribed it. We can even go a step further and imagine a scenario in which no agent produces the painting, no human agent that is: a robust, sentient computer system that generates new paintings every other week, stretches canvases on its own, and ships each painting out whenever there’s an invitation to exhibit at a gallery or museum. Say this computer-painter goes by the name “Asimo.” Let’s also say that Asimo operates immaculately over time; the occasional maintenance check, or any other instance of human intervention or micro-management is completely unnecessary. How can Graw’s definition accommodate and make sense of Asimo?

We can imagine Graw responding to this in a couple of different ways. She might point out that, regardless of how far removed (or in this case, non-existent) a human agent is from production, it will still take a human agent to have initiated production, e.g., to have programmed the actual computer and to have gotten things running, at least initially. Even if we modified the scenario, such that Asimo was to build another computer-painter, Omisa, and Omisa was to build another computer-painter, and so on, to the point that only computer-painters programmed other computer-painters who programmed other computer-painters, Graw might still suggest that the indexical effect can be causally traced back down the chain to some sort of human agent—the human programmer. Thus, in such a case, indexicality is maintained, if only through a series of computer surrogates.

Graw might also argue that regardless of the fact that the paintings are produced entirely by Asimo, the painterly effects, the gestures, the choices of color, and so on, nonetheless might still be experienced as highly personalized signs, despite the fact that the viewer may be unaware they were produced by a computer, and that therefore, somehow, their experience of them has been deceptively had (as the paintings were produced by computer-painter Asimo, rather than an actual human-painter).

Or, more reasonably thought, perhaps the experience isn’t considered deceptive or misleading at all. Maybe Graw thinks we ought to be receptive to the idea that the computer’s paintings do in fact constitute a highly personalized semiotic activity. Perhaps highly personalized semiotic activity should extend to sentient computers and machines, rather than strictly obtaining qua humans, or human surrogates.

Either way, it seems that (1) how the painting is produced and (2) whom the painting is produced by, will not play a particularly relevant role for Graw; an experience will qualify as constituting a highly personalized semiotic activity in each and every counterfactual scenario. Even if Asimo were to produce paintings with no gestural marks, or any sort of mark that would evidence the computer-painter’s hand—flawlessly printed, digitally-produced, monochrome paintings, say—we might find a moment of the personal in them someway, somehow. Maybe we notice flecks of dust on the surface, suggesting long-time studio neglect, or maybe, as Graw posits, just unadulterated decision making (e.g., choice of color, canvas, size, etc.) is sufficient enough to establish painterly autonomy and therefore qualify it as constituting a form of production of signs that is experienced as a highly personalized semiotic activity.


2. Fatal Flaws

Graw’s definition has two major flaws. The first is that it is overgenerative: it successfully operates within other categories of art, ones that are, in particular, not painting.

We might think of Marcel Duchamp’s readymade sculpture Bicycle Wheel (1913)—its intermittently-spun wheel an activity the artist compares to the experience of watching flames dance around in a fireplace; Felix Gonzalez-Torres’s 1991 work Untitled (Perfect Lovers) which features two identically ticking clocks slowly losing synchronicity over time, an allegory of Torres’s experience of slowly losing a lover to AIDS; or even one of Andy Goldsworthy’s rock assemblages, as qualifying as a form of production of signs that is experienced as a highly personalized semiotic activity.

One could reasonably argue that many non-painting artworks (if not all artworks) may be experienced as a highly personalized semiotic activity; prima facie nothing stands in the way of a work of art strictly disqualifying from this fact. For example, consider the various objects that comprise the above works. These objects act as signs that point to the artist’s latent presence simply in virtue of their epistemic status as artworks. That is, knowing that the above objects are artworks allows one to attribute authorship to them; it is to understand them as being produced by someone—and not immaculately conceived or created ex nihilo. It’s in this way that non-painting works of art may index their producer’s presence in the same way Graw claims paintings do.

Now, it could also be argued that the aforementioned works are experienced as a highly personalized semiotic activity only after a viewer has been informed of how they came to be and what they are about. That is, on first blush, one may not get a highly personalized reading, but, on a second, contextualized glance, one does. Here Graw might contrast such kinds of works with painting, which she might think passes as highly personalized on a first read. No contextual description necessary.

While this line of thought presumes that there is some highly personalized property, call it P, that painting and only painting possesses, it is unable to explain what features of painting ground P, where those features are located, or how they are accessed and derived. We can then ask: if P is not grounded in anything medium-specific, that is, if we understand painting to be a medium-unspecific art form not defined by any medium-specific properties, what else might P be grounded by? Perhaps, it is the property of being an artwork, in which case, it will be quickly pointed out that such a property isn’t painting-specific at all, but one that is inclusive of all artworks.

But, say P is not grounded in the property being an artwork, it then must be grounded in something else—and it can’t be the property of being a painting for obvious reasons of circularity. In fact, our property predicate can’t incorporate any painting verbiage at all, without being viciously circular and question-begging. Therefore, if P can’t be grounded in a painting-specific property (as opposed to medium-specific) then P declines from qualifying as a painting-specific property.

This is the first fatal flaw of Graw’s definition: the scope of what one may take to be highly personalized is far too wide; the criterion is simply overly inclusive.

While Graw’s definition is liberal enough to accommodate a surplus of criteria for why something might qualify as being highly personalized, by permitting this, no distinction between varieties and flavors of the personal can be sketched. Thus, her definition refuses to remain relevant to the aesthetic category in need of specific characterization—painting.11 Somewhat strangely, Graw is aware of this and even concedes that the case can be made for sculpture, video, and other aesthetic categories, but (and here is the argument): only to a lesser extent (2012, p. 52).12 It’s by this metric, “to a lesser extent,” that Graw’s defense crucially depends on.

It’s obvious that this response falls short. What exactly explains painting’s privileged highly personalized semiotic status in comparison to other categories of art? Graw’s answer here is fallacious, and her defense is mounted on an appeal to art-historical traditions and opinions which attest to painting’s “intrinsically intellectually demanding” (Leon Battista Alberti, 1453), “inherently and infinitely subjective” (Hegel, 1826), “subject-like” status (Louis Marin, Hubert Damisch), which evidently should be enough citations to coerce us (2012, p. 52 - 54). However, I believe nothing convincing can be found in these mentions: the intuitions of art theorists past don’t hold their weight simply by citation simpliciter. And while perhaps there are, historically speaking, shared intuitions over painting’s unique semiotic status, this should by no means ipso facto determine what ours are at present.

By way of a quick and crude example, consider the theoretical space substance, luminiferous aether, which was thought to exist by physicists and philosophers before the late 1800s or so. Aether, although undetectable, was introduced to explain the wave-like properties light seemed to exude in various experiments at the time. Like Graw, we too could point to more than a handful of thinkers from history who might have confidently asserted, or at least speculated, the existence of aether. And maybe this would be enough for us to think aether was theoretically relevant. But, post-1800, aether was rendered redundant; today we have good reason to reject the existence of such a substance (our theory of relativity, in particular, offers somewhat solid theoretical refutation).

Of course, the convictions of thinkers past should by no means be so hastily discarded or disregarded in such careless swoops. We should, however, find their theories only as good as the supporting text, and weighed accordingly against current data and intuitions. Hegel might not have even thought Rauschenberg’s Combine paintings could suffice as real, true-to-definition, paintings, let alone counted them as “inherently and infinitely subjective.” Why then should his intuitions, whose reach is restricted to a certain era of painting, be relied upon to do significant theoretical work? And perhaps painting could, at one point in time, have been accurately characterized as an “intrinsically intellectually demanding” activity, as Alberti considers it (specifically when techniques of geometry had to be acquired in order to faithfully depict reality), but today it would hardly be thought of in such terms.

While Graw sees these historical sentiments as evidence attesting to the close bond between product and person that painting has over time cultivated (2011, p. 116), I see them as just the opposite: unreliable sources; testimonials that prescribe either arbitrarily-stipulated, or historically moot properties to our contemporary notions of painting. But let’s say we were to grant them plausibility—we are still left with the question: why exactly is painting so semiotically privileged? So far this question has yet to be sufficiently answered by anyone, Graw included.

In a brief reply to Graw, art historian Peter Greimer queries film and video’s exclusion from her account of indexicality: why are film and video works unable to visibly capture the indexes of the producer in the same way that painting does? (2012, p. 60). Is the camera and editing work typical to film production not considered a highly considered and selective creative process, capable of conveying style and authorship?

In response to Greimer, Graw tries for a slightly expanded extrapolation:
“...in painting, this bond between product and person is especially unbreakable, as its signs refer to the producer consistently and not only selectively, like in film” (2012, p. 62). But, what is meant by “consistently” and “selectively?” Admittedly, it is highly unclear to me what we should take Graw to mean in this reply. Nevertheless, in an effort to elucidate what Graw has in mind, I will try to (briefly) formulate a charitable exegetical interpretation.

It might be construed that painting, unlike sculpture, video, or performance, is highly personalized by virtue of its conventionally inherited “aesthetic explicitness.” By this unfortunate, but temporary, jargon, I mean that painting is the most obvious way to make, or attempt to make art in the western world. To make a painting is, in a way, to epistemologically (as well as ontologically) guarantee its status as an artwork, regardless of its quality, i.e., whether it’s a really bad painting or a really good one. On the other hand, a readymade sculpture, a mere ordinary object, or a video of a clown being tortured,13 for instance, is perhaps less clear, again epistemically speaking—at least according to those outside the world of art (although, arguably, this is not so much the case anymore). Thus, painting’s privileged, highly personalized, semiotic status is distinguished from other aesthetic categories because of its derivation out of said normative conventions.

Returning to Graw’s response then, in film (in contrast to painting) it might not be explicitly apparent what signs count as traces of the producer. In other words, it is inconsistent, or only selectively apparent (to use Graw’s words), how much creative autonomy is exerted, or should be granted to the producer of the work. In comparison, in painting, this creative autonomy is brazenly splayed across a familiar rectangle. In film, a different kind of rectangle, there are more variables—perhaps less things one has creative control over. In painting, these variables are more manageable, as they are quite literally, physically contained.

Now, if my exegesis is correct, and the above is indeed within the vicinity of what Graw has in mind in her reply to Greimer, then I believe her to be wrong.

Imagine the case of a panicked lost dog who accidentally enters a painter’s studio, spills over various cans of paint, and darts back and forth across a freshly stretched linen canvas (laid out on the studio floor, wall-to-wall, floorboard-to-floorboard) as it frantically looks for an exit or its owner. The dog has run so many times back-and-forth across the canvas that its paw-prints have become indiscernible; they have now aggregated into sweeping, gestural splotches of color (call them human-like even). The dog eventually finds its way out, and the painter returns later that evening suspecting nothing. In fact, the painter thinks it was he himself who must have spilled the paint, and even more, impressed by this transcendental studio mishap, debuts the painting at his solo show later that month as-is.14

Will Graw say that the signs here refer consistently (again to use Graw’s language) to the producer, the painter, even though the work was entirely the result of an accident, produced accidentally by the dog as well? The counterfactual is slightly similar to the case of Asimo (in that it puts pressure on how one goes about assigning creative autonomy to an artwork), and so I anticipate Graw still standing her ground. For as I’ve shown, Graw’s definition would likely accommodate this accidental creation—attributing authorship to the artist, the dog, or to both—as accidental creations are still experienced as highly personalized regardless of their status of intentionality. That is, once more, that the indexical effect will obtain regardless of how a work is produced or by whom the work is produced.

But, if this painting-specific indexicality is grounded in an agent’s creative exertion, and if this creative exertion is subject to variability—as in the case of the example involving the dog above—then why be selective in how this creative exertion is evaluated across various categories of art? If, for example, film demonstrates the same type of variability that painting does, why does Graw’s indexical effect not obtain for it, as Greimer rightly observes? We return, once again, to asking why and how Graw’s account of indexicality remains applicable solely to painting, and not to sculpture, film, performance, and so on.

More seriously, if we can admit that this Grawian “unbreakable bond,” between painting and painter is “open” (“open” in the sense that it will be null and void whether a creative act was exerted intentionally, accidentally, directly, indirectly, and so on, for Graw’s indexical effect to obtain), and even further, that a viewer may come to believe and experience this unbreakable bond, why, we may ask, is this bond uniquely tolerated by painting beyond arbitrary assignment? I don’t believe it is. If Graw cannot answer this question, her account remains entirely unmotivated.

Here, then, we arrive at the second fatal flaw of Graw’s definition: If fortuitous circumstance is a tolerable condition in granting the sign/producer consistency criterion that Graw attributes specifically—and solely—to painting, then there should be, in principal, absolutely nothing preventing film, or other categories of art from qualifying as well. Any restrictions thus imposed will be ad-hoc.

The bond between product and person is, contra Graw, not especially unbreakable for painting—it is in fact quite medium-encompassing. There is nothing, in the Grawian sense, personally special about painting.


  1. In the art world at least, although I do not speak for everyone.
  2. See David Joselit (2009).
  3. Graw opens up her text with this very remark: “In the following I will first try to develop a medium-unspecific notion of painting that is nevertheless able to capture its residual distinctness even under conditions that led to its diffuse boundaries.” (2012, p. 1).
  4. To be fair, Graw never officially calls this a definition, however, it’s quite clear from her texts (Graw, 2011, 2012) that it is to be understood and applied as such.
  5. I say “Peircean-like” because Graw presents a modified version of Peirce’s account. In a response to art historian Peter Geimer she stresses this point, specifically that she is “not adopting the Peircean model one-to-one.” (2012, p. 61–62)
  6. In the spirit of anthropologist Alfred Gell’s theory of aesthetic agency (1998).
  7. This obviously has both psychological and economic interpretations, but we are not interested in exploring those interpretations here (as interesting as they might be).
  8. Our “living being” should be one that, at the least, possesses higher-order cognition, e.g., a person, a certain kind of animal, etc. (Although since this is certainly open to dispute, one may simplify things for the time being by restricting the scope to persons.)
  9. Some might want to compare these notions to Walter Benjamin’s concept of “aura.” However, I’m not qualified to gauge how helpful or informative such a comparison would be.
  10. Note that I use highly personalized semiotic activity or highly personalized as shorthand for Graw’s definition here and elsewhere throughout the text.
  11. At this juncture, some might think this is very well Graw’s point: that her definition isn’t meant to stringently characterize painting in any medium-specific way. But to those who are of this mindset, here is a question: what about painting sanctifies Graw’s thinking that there is this unique, highly personalized, semiotic trait specific to it, if there is not something medium-specific about it?
  12. Here is the full quotation: “Isn’t sculpture marked by a similar kind of indexicality and it therefore also suggests that it is a quasi-person? Yes, it does, but to a lesser degree. Only painting has many historical arguments pointing to its subject-like power—arguments that I believe do reach into our present.” (2012, p. 52 my, italics)
  13. See Bruce Nauman’s Clown Torture (1987).
  14. A version of this thought experiment was developed out of a discussion I had with artist Kat Schneider, April 6, 2015.


Isabelle Graw. “The Value of Painting: Notes on Unspecificity, Indexicality, and Highly    Valuable Quasi-Persons,” in Thinking through Painting: Reflexivity and Agency beyond the Canvas, eds. Isabelle Graw, Daniel Birnbaum, and Nikolaus Hirsch (Frankfurt am Main: Sternberg Press, 2012).

———. “Knowledge of Painting: Notes on thinking images, and the person in the product.” Texte Zur Kunst, June, 2011, pp. 114-124.

David Joselit. “Painting Beside Itself.” October 130 (2009): 125-134.

Donald Judd. “Specific Objects,” 1964. Arts Yearbook 8 (1965): 94; reprinted in Thomas Kellein, Donald Judd: Early Works 1955-1968 (exh. cat.). New York: D.A.P., 2002.

Rosalind E. Krauss. Under Blue Cup. Cambridge, MA: The MIT Press, 2011.

———. A Voyage on the North Sea. Art in the Age of the Post-Medium Condition. London: Thames Hudson, 1999.

Jan Verwoert. “Why Are Conceptual Artists Painting Again? Because They Think It’s a Good Idea.” Afterall Autumn/Winter 12 (2005): 7-16.


Louis Doulas

LOUIS DOULAS is a Master’s candidate in Philosophy at Brandeis University.