6 Collingwood, Emotion, and Computer Art
Margaret A. Boden
6.1 Introduction
Philosophical aesthetics and computer art seem to be miles away from each other. Certainly, the mainstream philosophical literature does not address computer art more than perfunctorily if at all. It is sometimes said, almost as an aside, that “computer art” is a contradiction in terms (e.g., O’Hear 1995). But because there is no universally agreed definition of art in the philosophical community, it is not obvious that that is so. Moreover, because there are distinct varieties of computer art (see chapter 2), it may be that some have more claim on this honorific title than others.
One implication of the lack of an agreed definition of art, of course, is that there can be no knockdown argument about the status of (any type of) computer art as art that addresses all the many definitions that have been offered. This chapter examines just one of these accounts of art and asks what its implications are for the various forms of computer art.
Specifically, it considers the philosophical aesthetics of Robin Collingwood (1889–1943).
6.2 Why Pick Collingwood?
The question Why pick Collingwood? has two answers. The first is simply, Why not? Collingwood is recognized as “one of the twentieth century’s few outstanding philosophers of art” (Ridley 1998, 1). Only fifty years ago, he was “probably the most widely read and influential aesthetician to have written in English since Addison, Hutcheson, and Hume” (Kemp 2009). Given that he is such an important voice in philosophical aesthetics, it is of interest to ask how his views relate to any kind of art.
To be sure, his views are less popular today than they were half a century ago. Indeed, parts of his work are now nearly universally “scorned” (Guyer 2003, 45). Nevertheless, he still has his champions. In 2003, for instance, Aaron Ridley spoke of Collingwood’s “wonderful” book Principles of Art (1938), while admitting that it is also “wonderfully uneven” (Ridley 2003, 222). Critics who have “scorned” much of his work acknowledge other parts of it as philosophically “valuable” (Guyer 2003, 45). And more generally, current writings on aesthetics abound with passing references to him. In short, Collingwood is still a highly respected name in this area.
However, so are others. For example, Arthur Danto is often cited in philosophical aesthetics. He defined art in relation to the institutions of the art world—something Collingwood would have regarded as essentially irrelevant (Danto 1964, 1981; Carroll 1993, 80). In so doing, he raised many of the issues explored by Ernest Edmonds in chapter 8 of this volume. So why not choose him?
Well, most of the people who are skeptical about the very idea of computer art—and in my experience, that means most people—do not base their skepticism in Danto’s approach. Far from it: if the art world were to become more accepting of computer art, their doubts would remain. If a major collector such as Charles Saatchi were to invest in C-art, it would (on Danto’s account) become “art” almost overnight. But the general public would very likely not be persuaded.
Danto’s institutional theory was developed in part to oppose the skepticism of mid-twentieth-century audiences who, on first encountering much pop art and Conceptual art, reacted in outrage, saying “That’s not art!” His answer, in effect, was “But the art world accepts it as such, and that settles the matter.” Their riposte would generally be “So much the worse for the art world!” They would probably be even more dismissive of computer art than they are of the provocative urinals, Brillo boxes, and buried kilometer rods—and now the diamond-encrusted skulls—that inhabit the computer-free zones in modern art (Boden 2007b).
The reason for this is that people’s unwillingness to countenance the possibility of computer art, as real art, is usually based in a strong intuition that goes back well over two thousand years—namely, that art somehow involves emotion, something that, they believe, computers could not possibly have.
We need not challenge that belief here: let us grant that computers do not have emotions. Which is not to say that they cannot be used to model emotion in ways that are psychologically illuminating (Boden 2006, 7.i.f). But we do need to question the “how” in that “somehow.”
This gives us the second answer to the question “Why pick Collingwood?” As we will see, Collingwood put emotion at the heart of art even more firmly, even more exclusively, than most philosophers do. It follows that his theory is prima facie an especially strong challenge to the very idea of computer art. In responding to that challenge, the relationship of emotion to C-art must be clarified.
Those philosophers who put emotion in pride of place as a criterion for genuine art do so in several different ways. For instance, Clive Bell (1914, 113), whose writing was hugely influential in the rise of abstract art, insisted that “the starting-point for all systems of aesthetics must be the personal experience of a peculiar emotion”—one aroused, and justified, only by “significant form,” which is therefore the “essence” of all visual art. Although this emotion is experienced by the person, its intentional object, on Bell’s view, is strictly impersonal. The aesthetic emotion responds not to any human predicament or story, whether joyous or loving or fearful, but only to abstract form. Other (expressionist) philosophers claim that art involves the expression of some more personal emotion and that art appreciation is grounded in the experience of a similar emotion by the audience.
Both these positions, along with all other forms of expressionist aesthetics, were rejected by the modernists and in particular by the conceptual artists of the 1960s (Boden 2007b). One of their leaders contemptuously dismissed claims that art lovers are driven by “the expectation of an emotional kick” (LeWitt 1967, 79).
But art lovers in general, as Sol LeWitt realized full well, are more sympathetic to expressionism. They may doubt whether emotion is actually the essence of art—being perfectly willing, for instance, to accept the affect-free drawings of Maurits Escher as genuine art. But they readily accept that art and emotion are typically closely linked. One example of this attitude is mentioned in chapter 4: the artificial intelligence (AI) scientist Douglas Hofstadter, who was surprised by the seemingly emotional content of the music composed by David Cope’s Emmy program—“seemingly” emotional because, as explained in chapter 3, Hofstadter was projecting emotion onto the music rather than finding it within it.
Collingwood was a hugely influential expressionist. It is no accident that his position began to be “scorned” with the midcentury arrival of conceptual art. But his version of the thesis was unusual. One of the most prominent differences from other expressionist aestheticians, such as Leo Tolstoy (1899), was his unrelenting particularism. As we will see, computer art fails to satisfy his concept of art, even though it could match some of the alternative emotion-based accounts offered by other expressionist philosophers, largely because he defined it in that way.
In this chapter, Collingwood is not regarded as a philosophical guru to be slavishly followed on all points. Rather, he is treated as a worthy intellectual sparring partner. His ideas, whether acceptable or not, can help us explore the relations between computer art and the more traditional denizens of our galleries and concert halls.
6.3 Collingwood’s Philosophy of Art
Collingwood offered two quite different philosophies of art, of which only the second, developed in the 1930s, concerns us here. It is that second account, so hugely influential, whose footsteps can be seen in many discussions of art today. That is not because it was the more immediately plausible. To the contrary, it was highly surprising, even challenging.
His earlier writing on aesthetics had been relatively unexceptional (Collingwood 1925). It was based on a long-familiar idea: the appreciation of beauty. Art, he said in the early 1920s, is “the special activity by which we apprehend beauty” (9). He insisted that “the awareness of beauty is at once the starting-point and the culmination, the presupposition and the end, of all art,” and “it is simply an enlargement and a sharpening of the [artist’s] awareness [of beauty] that constitute, either for him or for anyone else, the value of the picture [or other artwork] when it is done” (1925, 7).
The imagination, he said, enables the appreciation of beauty (“the beautiful is neither more nor less than the imagined” [1925, 19]). And it is crucial to the practice of art, which is “the primary and fundamental activity of the mind, the original soil out of which all other activities grow” (1925, 14). But not all beauty is created by art. He saw Kant’s notion of the sublime as the “elementary” form of the beautiful: it is “beauty which forces itself upon our mind, beauty which strikes us as it were against our will and in spite of ourselves, beauty which we accept passively and have not discovered” (1925, 35). Human artistry, to be sure, is far from passive. But the most basic aesthetic response occurs “against our will and in spite of ourselves.”
The emphasis was less on the artist as originator than on our aesthetic response to the work of art itself. Indeed, there need not be any artist. Beauty could be appreciated in manmade objects designed for nonartistic purposes—or, of course, in nature.
By contrast, Collingwood’s (1938) later aesthetics focused on how the work of art reflects certain aspects of the human artist. And this theory was unorthodox, even counterintuitive, in various ways.
Beauty, now, was given no special importance: if present, it could be relished, but it was not necessary in order to give art its aesthetic worth. By implication, significant form—the modernist notion of beauty (which had gained currency since its early century introduction by Bell and by the painter Roger Fry)—was an optional extra, not (as they had said) the very essence of art.
Further unorthodox claims made by Collingwood included his rejection of representation as a criterion of art; what he meant by emotional expression; what he identified as the work of art; and his distinctions between art and craft, entertainment, and propaganda. Each of these, as we will see, is relevant to the interpretation of computer art.
His insistence that art is not a luxury but is crucial for a humane and civilized culture was only slightly less unusual. What he meant by this seemingly trite claim was idiosyncratic. His view was that art has a hugely important practical effect. Emotional self-knowledge, he said (for reasons explained later), is necessarily provided (to the artist) by good art and is constructed also by the audience who understands it. And such self-knowledge is a necessary, but not a sufficient, condition for a healthy society.
Specifically, the fascist politics sweeping across Europe at the time thrived, he said, largely because the public could not recognize the “corruption of consciousness” involved in what was presented as art but was really emotively manipulative propaganda. Lacking emotional self-knowledge (provided only by art proper), people could not recognize or admit—still less criticize and reject—the emotions aroused in them by such effective propaganda. These emotions were not merely (often) socially debased but also crude—that is, lazily unexamined and unchecked. As he put it, “I call this the ‘corruption’ of consciousness; because consciousness permits itself to be bribed or corrupted in the discharge of its function, being distracted from a formidable task towards an easier one” (1938, 217).
Similarly, modern advertising and mass entertainment rely on the skillful use of popular media to arouse emotions that are not examined and so provide no advance in self-knowledge. Thus, a society that is awash with emotions may, in the absence of art, direct these emotions inappropriately. He even went so far as to say that “bad art” is the root of all evil: “the true radix malorum” (1938, 285).
None of this implied that artists are a superior class, to be entrusted with the responsibility of saving society from itself. On the contrary, said Collingwood, “the effort towards expression of emotions, the effort to overcome corruption of consciousness, is an effort that has to be made not by specialists only but by everyone who uses language, whenever he uses it. Every utterance and every gesture that each one of us makes is a work of art” (1938, 285; italics added).
His emphasis on self-knowledge and emotional honesty, in both artist and Everyman, explains why Collingwood saw bad art as so destructive. By bad art, he did not mean a clumsy cut with the chisel or a carelessly wielded paintbrush or even a sycophantic portrait or a badly represented landscape. Rather, a bad work of art is “an activity in which the agent tries to express a given emotion but fails” (1938, 282). As for trite popular novels and movies or attention-catching advertisements, these are not bad art so much as quasi art. They do not count as art at all, for they promote “organized and commercialized day-dreaming” rather than the true expression of emotions (1938, 138).
What is it, then, to express an emotion? For Collingwood, this is not a matter of first having an emotion fully formed and then expressing it faithfully to transmit it to the audience. Rather, to express the emotion is to construct it, to identify it, and in so doing to specify and form it. In his words, “Until a man has expressed his emotion, he does not yet know what emotion it is” (1938, 111).
This may seem absurd: surely I can know that I feel anger or grief or joy without expressing it to myself or to anyone else? Collingwood would not disagree. But he would say that this indefinite feeling is preconscious rather than conscious. The particular emotion that you feel, the specific nature of that instance of anger or grief or joy, can be identified and recognized only after it has been developed and sharpened by being consciously, and conceptually, expressed. That is, a particularized emotion is not pure feeling. Rather, it is an intellectual construction—“a feeling … dominated or domesticated by consciousness” (1938, 217).
The act of consciously constructing emotion, rather than merely suffering it or even communicating it, is essential in Collingwood’s aesthetics. So he compared artistic and emotional expression to the struggle to express what one is thinking (1938, 267–269), which, as we all know from personal experience, helps construct the thought in some newly definite form rather than simply conveying a preexisting conception.
Perhaps the clearest statement of Collingwood’s particularism was this: “The artist proper is a person who, grappling with the problem of expressing a certain emotion, says ‘I want to get this clear.’ It is no use to him to get something else clear, however like it this other thing may be. Nothing will serve as a substitute. He does not want a thing of a certain kind, he wants a certain thing” (1938, 114). It followed, he said, that everyone who reads literature as psychology, as depicting “the feelings of women, or bus-drivers, or homosexuals …, necessarily misunderstands every real work of art with which he comes into contact, and takes for good art, with infallible precision, what is not art at all” (1938, 114–115).
Once the artist’s emotion has been expressed in this way, it can in principle be communicated to another person. But again, this communication is not what is often meant by that term. Many people claim that art enables emotion of a certain type to be transmitted from one person to another. It is almost as though the medium (text, painting, musical instrument, and so on) carries the emotion as a carrier-pigeon, picking it up from the mind of the artist and depositing it in the mind of the audience. For Collingwood, this is impossible. Or rather, it is possible, but only in the case of nonart, such as entertainment (including, today, TV soap operas) or political propaganda. In art proper, the audience constructs the emotion afresh in their mind, much as the artist did in her or his, and it is the equivalent, highly particular, emotion that is involved.
Several things follow from this. One is that “there is no distinction of kind between artist and audience” (1938, 119). Collingwood endorsed Samuel Taylor Coleridge’s remark that “we know a man for a poet by the fact that he makes us poets” (1938, 118). Granted, the artist is the originator. And the artist is the person who uses hard-won skills to produce the art object: a point taken up at length in chapter 4. But the mental work—the construction of a particular emotion—done by the audience is essentially the same as what went on in the artist’s mind.
As Collingwood put it, “Though both [poet and audience] do exactly the same thing, namely express this particular emotion in those particular words, the poet is a man who can solve for himself the problem of expressing it, whereas the audience can express it only when the poet has shown them how” (1938, 119). In other words, Everyman an artist: “The poet is not singular either in his having that emotion or in his power of expressing it; he is singular in his ability to take the initiative in expressing what all feel, and all can express” (1938, 119).
The artist’s experience is richer too. “If you want to get more out of an experience, you must put more into it.” Painting a picture will necessarily be a richer experience than merely looking at it or looking at its subject in the natural world: “The painter puts a great deal more into his experience of the subject than a man who merely looks at it; he puts into it, in addition, the whole consciously performed activity of painting it; what he gets out of it, therefore is proportionately more” (1938, 308). In sum, “[He ‘records’ in his picture] not the experience of looking at the subject without painting it, but the far richer and in some ways very different experience of looking at it and painting it together.”
Another implication—on which Collingwood insisted repeatedly—was that the “work of art” is not a physical object, such as a painting or sculpture, or a physical phenomenon, such as a series of visual images or sounds. Nor is it a public performance, like ballet or street theater. It is not even an abstract structure recorded in some formal notation, like a musical score or written text. Rather, the work of art is a mental entity: the (fully formed, identified, subtly expressed) emotion in the artist’s mind—together with its psychological twin in the mind of the audience. The artist’s emotion, albeit mental, is not an essentially private matter. To the contrary, it is expressed—and mediated—by public “languages”: words, gestures, and culturally familiar artistic conventions.
In one sense, Collingwood regarded all these things, and everything else too, as mental constructs. He was committed to an Idealist metaphysics, developed in his previous writings. For present purposes, however, this fact can be ignored. His contrast between the constructive mental activity that is involved in understanding art and the absence of this in perceiving things without understanding was independent of his background Idealism (see Ridley 1998, 17–25).
That a work of art is not a physical thing or even an observable phenomenon may seem a strange idea. But remember the conceptual artists of the 1960s and their deliberate “dematerialization of the art object” (Lippard and Chandler 1968, 31; Lippard 1973; Boden 2007b). For them, it was the idea in the artist’s (and audience’s) mind that was the true work of art. There might not be any physical artifact associated with it, and if there was, it might be invisible (e.g., Walter de Maria’s buried kilometer rod). Moreover, other philosophers of aesthetics besides Collingwood have offered versions of mentalism. For instance, David Davies (2004, 146) argues that every work of art is an action—or in a sense, a performance. Specifically, it is an action on the artist’s part, which identifies a “focus of appreciation” whose vehicle may be, but need not be, a physical object. Even if it is, the work of art as such is the artist’s attention-directing action. So if Collingwood was unusual in being a mentalist, he is not unique.
Collingwood sometimes used the term “work of art” in its more familiar sense (e.g., 1938, 37), applied to a physical artifact—what one might call the art object (and what Danto calls the “vehicle”). But he did so as a sort of shorthand. The artist’s emotion is often expressed (particularized) with the help of paint on canvas, chiseled marble, or words on a page. On Collingwood’s view, however, the art object should not attract attention for its own sake: for its phenomenal qualities or beauty. It is an “objet d’art … only because of the relation in which it stands to the aesthetic experience which is the ‘work of art proper’ ” (ibid.). Its role is to function as a “language” that can prompt the culturally prepared audience to go through the same sort of constructive mental activity as the artist did when making it.
And just as what really matters about a book is not the printer’s ink on the paper or the beauty of the typeface but the meanings expressed by way of them, so the physical art object is not what art is really about. It is not the work of art itself.
By the same token, the medium chosen by the artist is artistically inessential. The artist may have found that a certain medium is best, or even in practice necessary, for expressing or constructing the emotions concerned. But in principle, and perhaps in fact, some other medium might do so just as well.
Yet another implication is that art, as such, has nothing to do with usefulness. What Collingwood meant by usefulness did not include the benign social-political effects of art previously mentioned. He granted that “a work of art may very well amuse, instruct, puzzle, exhort, and so forth, without ceasing to be art, and in these ways it may be very useful indeed.” Nevertheless, he insisted that “what makes it art is not the same as what makes it useful” (1938, 32).
Likewise, art has nothing to do with representation (mimesis), previously thought by many philosophers to be the core aim of art. A work of art may in fact be a successful representation. But for Collingwood, what makes it good art is not the same as what makes it a good representation.
And finally, art has nothing to do with skill (a counterintuitive notion that is challenged in chapters 4 and 10). Certainly, a work of art—or rather, an art object—may be fashioned with prodigious skill. But, he said, the skillfulness qualifies it as a superior work of craft, not of art. Skill in working in a particular medium may be essential in practice if an artist wishes to express a particular emotion clearly enough for others to be able to experience it too. Even skill in representation may be necessary if a painted portrait or landscape is to enable the audience to construct the same emotion as it did in the artist. But skill is not the same as art. Far from it: “Art proper, as the expression of emotion, differs sharply and obviously from any craft whose aim it is to arouse emotion” (1938, 113).
Collingwood saw this distinction as “sharp” and “obvious” (which in fact it is not; see Boden 2000) because he saw the aim of craft as very different from the aim of art. On the one hand, craft is not particularist in intent: “The end which a craft sets out to realize is always conceived in general terms, never individualized” (1938, 113). And, on the other hand, he said, craft (skill, technique) is applied to preordained ends.
These ends include making pots from lumps of clay or tables from planks of wood—for which, according to Collingwood, no highly skilled craftsman begins to work without having a clear idea of what it is he wants to make. Similarly, the craftsman may seek to provide magic, entertainment, or propaganda—all involving the foreseeable elicitation of a general type of emotion. But the hard work that is involved in clearly expressing a highly particular emotion is hard largely because its end is not foreseen. Only after it has been successfully expressed does the artist’s emotion exist as an identifiable particular experience. In that sense, art is not goal directed but open-ended.
Like Danto after him, Collingwood deliberately sought to include the newest art as well as (some of the) more familiar types. Right at the outset of his book he spoke of “a new drama, … a new poetry, and a new way of painting … [and] a new way of writing prose” (1938, 1). He had in mind rising artists such as “Mr. Joyce, Mr. Eliot, Miss Sitwell, or Miss Stein” (1938, 3). The question to be considered now is whether an even newer form of art, computer art, fits within his definition.
There is no short answer. There are diverse forms of computer art, some of which may be better suited to Collingwood’s approach than others are. Moreover, there are two ways of thinking about computer art, of any category. One is to think of the computer as an artistic medium; the other is to focus on its generative aspects. These points are explored in the next three sections.
6.4 Computers as an Artistic Medium
Artists tend not only to use their chosen medium but to celebrate it. A composer, for instance, glories in the range of acoustic possibilities offered by the piano or the clarinet as well as in the more abstract musical structures represented in the score. And a skilled pianist or clarinetist exhilarates in coaxing the chosen instrument to reach unexpected heights in performance. Computer artists too—and especially those engaged in computer-generated (CG) art as opposed to computer-assisted (CA) art—tend to celebrate their medium rather than merely using it.
As defined in chapter 2, CG-artworks result when a computer program is left to run by itself with minimal or zero interference from a human being; in computer-assisted, or CA-art, by contrast, the computer is used by the human artist as an aid—in principle, nonessential—in the art-making process. We see later that this distinction is not always clear-cut: some CA-art relies heavily on CG-processes.
The computer being the chosen medium for all C-art would undermine the possibility of true computer art if something about computers disqualifies them from being an artistic medium. That is especially so if one’s theory of aesthetics treats the medium as an essential aspect of the work of art.
The latter condition, although still common, is less widely accepted today than it was in Collingwood’s time. The conceptual artists, twenty years after his death, made a point of denying that the medium, and the artist’s skill in handling it, is important. For them, and for their current followers, art is about ideas, not artifacts. Nor did this condition apply to Collingwood himself. He too, as we have seen, identified the work of art as a mental construct. He regarded the medium in which it is embodied as contingent, not essential.
Yet even Collingwood had to allow that the chosen medium must be one that, as a matter of fact, is capable of helping the artist and audience construct the emotional experience in question. This matter of fact is largely cultural. There may be some physical reason why a putative medium is unable to communicate a certain emotion or even any emotion at all. For instance, paints radiating only ultraviolet light would be imperceptible to humans. The art galleries would be visited only by bees. And collages made of individual mercury droplets would lose their form if the droplets were able to touch each other. But a medium may be unsuitable, instead, because of its cultural associations.
Consider, for instance, Chris Ofili’s The Holy Virgin Mary, the portrait of a black Madonna that won London’s Turner Prize in 1998. Mary’s gown was blue, respecting a painterly convention familiar since the Renaissance. But there the traditional aspects ended. Quite apart from the unconventional (black African) physiognomy, the medium was most unusual. The picture was made with a mixture of oil paints and elephant dung, and the linen canvas bore some small resin-coated balls of dung.
This work of art created a scandal because of what the British audience saw as its scatological, and therefore blasphemous, associations. A similar reaction greeted it in New York, where Mayor Rudolph Giuliani wanted it removed from the walls of the Brooklyn Museum of Art. In the African environment that had inspired it, however, elephant dung is considered highly useful, even near sacred. What is more, it is sometimes used for painting. And in parts of India the dung sells for high prices, so the artist’s use of it could be seen as comparable to using gold leaf or precious stones on a medieval religious icon.
In brief, there is no intrinsic property that makes elephant dung physically unsuitable as an artistic medium. The scattered dung balls might have rotted away, to be sure, but they were protected by the resin. As a matter of cultural fact, however, it cannot be used to make works of art that will be widely accepted in the UK or United States.
Of course, Ofili knew very well what he was doing. He was not painting for an audience in Africa or India, and he fully intended to épater le bourgeois in the West. He even included some sexually expressive symbols in the painting: several tiny images of buttocks and female pudenda—after all, that is where baby Jesus came from. But those potentially offensive images were not what enraged the public. Perhaps most people did not look closely enough to realize what they were. It was the choice of dung as the artistic medium that caused the furor.
The public’s distaste for computers does not quite match its distaste for feces, elephantine or not. Artists aiming to épater those beleaguered bourgeois have sometimes used human feces too—sealed in ninety small cans in the case of Piero Manzoni’s Merda d’artista of 1961. But there is no doubt that many people are highly suspicious of computers and are especially averse to their use in what are normally regarded as quintessentially human contexts.
Art is a paradigm case of such a context. And the emotional aspect of art is seemingly the most human aspect of all. If people believe—which most people do—that computers cannot experience emotions and indeed can have nothing whatsoever to do with emotions, then they are likely to think that computers cannot be used to communicate emotions. And for anyone who (like Collingwood) believes that emotion is the very essence of art, computers will probably appear to be utterly unsuitable as an artistic medium.
On that view, what the computer actually does is irrelevant. In addition, how it is actually used by the human artist is irrelevant. Likewise, the Madonna’s blue gown and even Ofili’s provocative mini-images were deemed irrelevant by the people scandalized by the medium chosen for his work. No specific C-art project needs to be examined. None merits examination, for “C-art” is a contradiction in terms. Both computer-generated and computer-assisted art are to be dismissed as frauds. Simply because they use computers as their medium, neither CG-art nor CA-art is really art at all.
This position is not rationally grounded. Oil paints cannot feel wonder and humility or marble, grief. But these media can nevertheless be used to fashion an awe-inspiring Annunciation or a hugely affecting Pieta. So the mere fact that computers cannot feel emotions does not rule them out as an artistic medium. What, if anything, rules them out as a medium is the widespread cultural prejudice that makes it impossible for most people to experience emotions (other than bemusement and frustration) when presented with computer art. This prejudice can even prevent people from being presented with computer art: one music critic wrote a savage “review” of a concert of CG-music even though he admitted that it had not yet happened and that he had no intention of turning up when it did (Cope 2006, 345).
If such prejudice is to amount to anything more than an unthinking suspicion of the medium, some other reason must be found by those who wish to reject computer art. In other words, how the medium is actually used and what the computer actually does must become the focus of attention.
6.5 Duplicability, Open-endedness, and Skill
Several of Collingwood’s criteria for art are prima facie problematic with respect to computer art. In this section and the next, however, we see that most of these criteria can be satisfied, in principle or (often) in practice, by this novel genre. Only one is strictly incompatible with C-art, or rather, with an important subclass of C-art.
The first such criterion is his mentalism. We see in section 6.3 that a work of art, for Collingwood, is not a physical thing. If that is so, then the images and sounds presented as C-art by C-artists are not really art at all. However, no C-artist is going to lose sleep on that account. By that criterion, even the Mona Lisa or Michelangelo’s David is not a work of art.
The work of art, Collingwood insisted, is the mental construct in the artist’s (and viewer’s) mind that expresses the experience concerned. Or rather, it is the relevant mental construction in the artist’s and multiple viewers’ minds. There are thus countless near-equivalent works of art associated with Leonardo’s or Michelangelo’s masterpiece—and with every other physical artifact commonly regarded as a work of art.
I say near equivalent rather than fully equivalent because each one is freshly constructed by the individual human mind concerned. Despite Collingwood’s extreme particularism—“ ‘I want to get this clear,’ not something else, however like it this other thing may be”—it is surely unrealistic to suppose that the relevant experiences are exactly alike in every respect.
It would seem to follow that the mere duplicability of a computer artwork—or for that matter, of a traditional print (see section 6.6)—should not count against its status as a work of art. On the one hand, there are always many near-equivalent (mental) works of art. And on the other hand, a (physical) print—again, for Collingwood—has no such status anyway. So a certain print being unique can give it rarity value as an object within the art market but does not enhance its value as art.
By contrast, for those people who assume that some works of art are physical things, duplication often does undermine the work’s claim to be taken as art. This does not apply to Walter Benjamin, who rejected uniqueness as elitist: he welcomed technologies of art reproduction, such as still photography and film, as democratically progressive (Benjamin 1968). A print, lithograph, or etching, no matter how much its skill is admired, is usually less revered than a painting. Even if only one example of it actually exists (perhaps because only one was printed or because all the others have been lost), the mere potential for duplication can compromise its perceived status as art.
With respect to computer art (especially CG-art), this certainly happens. That is, critics often downgrade it precisely because they feel that, as computer output, it can be duplicated or nearly matched indefinitely. That is one of the skeptical reactions that greeted David Cope’s Emmy program, which composed highly plausible music modeled on the styles of human composers from Mozart to Joplin (Cope 2001, 2006). And it is one of the reasons Cope decided to destroy the musical database that he had taken twenty-five years to build, thus making Emmy’s oeuvre complete—as those of deceased human composers are (Cope 2006, 364; Boden 2007a).
In short, duplicability (among other things) was preventing people—such as the concert-avoiding “reviewer” mentioned earlier—from treating Emmy’s music as music and from taking Cope’s C-art seriously as art. Were one to accept Collingwood’s mentalism, however, that reaction would be ill based.
What of Collingwood’s insistence that true art is open-ended? If this means that it is unpredictable, that the end point cannot be foreseen, then CG-art (and even most CA-art) is included. To be sure, one can usually (although not quite always; see later discussion) predict, in general terms, what type of CG-result will ensue. But unpredictability in detail is par for the computer course.
In fact, utter unpredictability is not a feature of human art either. In other words, one must question Collingwood’s curious claim that “no artist, so far as he is an artist proper, can set out to write a comedy, a tragedy, an elegy, or the like. So far as he is an artist proper, he is just as likely to write any one of these as any other” (1938, 116). This has some plausibility when applied to a poet laureate, committed by his office—not by his royalist sentiments—to write an elegy for a dead king or a prothalamion for a prince’s wedding. It has some plausibility, too, in cases in which an artist is specially commissioned to produce a work of art that otherwise she or he would not have undertaken. In general, however, the claim is highly dubious. A poet enthused by a daughter’s pending marriage could surely “set out to write a [prothalamion]” in celebration and could identify or express his own particularized emotion in the process.
It follows that the mere fact that no computer program is “as likely to [generate any one result] as any other” does not mean that it is not unpredictable in more relevant ways.
The best-known case of CG-art, Harold Cohen’s AARON program, can generate indefinitely many unique drawings and colored prints or electronic displays (Cohen 1995, 2002; Cohen, Lugo, and Moura 2007). The works produced by the most recent version of the program “teeter on the edge between abstraction and representation,” allowing color to function as “the primary organizing principle, whilst maintaining a link to the external world” (Cohen and Jacobson 2008). Earlier works were more obviously representational, and AARON’s competence in mimesis improved considerably as the program was developed over the years. So did its competence as a colorist. Indeed, Cohen (personal communication) has said, referring to a fairly recent version, “I am a first-rate colorist. AARON is a world-class colorist.”
Quite apart from their interest to people who are knowledgeable about CG-art and their novelty value (as examples of computer art) to those who are not, AARON’s drawings are both realistic enough and attractive enough to be exhibited around the world—in the Tate Gallery, for example. Collingwood would not be persuaded by that, of course. For him, as we have seen, success in representation (mimesis) is artistically irrelevant. Likewise, visual attractiveness and entertainment value do not add up to art.
More to the point, in this context, is that AARON’s pictures are individually unforeseen and unpredictable—even by Cohen himself. In that sense, the program is open-ended.
Certainly, the images generated by any given version of Cohen’s program (which he developed over forty years) are all drawn in the same style and depict the same general theme—for example, people with jungle foliage or acrobats with medicine balls or multicolored leaflike abstract designs. However, human artists too, even great artists, spend almost all their time in exploring a certain style. They vary the subject matter more than AARON does (although Claude Monet was temporarily obsessed by cathedrals or water lilies, and William Turner had a thing about sunrises—so much so that he lived in several houses on England’s eastern coast). But their general approach and also their personal signature (Boden 2010) vary only a little.
Besides such exploratory creativity, which leaves the style better understood but essentially unchanged, human artists sometimes (although rarely) engage in transformational creativity (Boden 2004, chaps. 3, 8, 12). This alters one or more aspects of the style, one or more dimensions of the relevant conceptual space, so that new structures or ideas can now be produced that were impossible before. The passage from impossibility to actuality surely counts as open-ended. And Collingwood, as we have seen, was as favorably impressed as anybody by the “new poetry” and the “new way of writing prose” found in the transformational work of “Mr. Joyce, Mr. Eliot, Miss Sitwell, or Miss Stein.”
Insofar as CG-art (as opposed to CG-artists) can produce stylistic transformations, it does so by means of evolutionary programming, or Evo-art. This was defined in chapter 2 as CG-art that is evolved by processes of random variation and selective reproduction that affect the art-generating program itself.
Whether Evo-art can produce truly fundamental changes is discussed in chapter 10. Here, let us just note that it sometimes appears to do so. This is especially likely if some of the changes in the program’s own rules are structurally radical—such as concatenating two whole image-generating miniprograms or hierarchically nesting one inside another and perhaps another (Sims 1991). Although the types of rule change are predetermined by the programmer, the individual changes that actually happen are random—much as biological mutations are. There is no question, then, of the resulting Evo-art being foreseen by the human artist.
Admittedly, the types of mutation allowed are sometimes deliberately limited, so as not to depart too far from the original style. For example, a series of point mutations, in which one numeral in a programmed instruction is substituted by another, generate changes that look like adventurous stylistic exploration rather than radical transformation (see Todd and Latham 1992). But even so, the human artists concerned avow that they could not possibly have imagined (still less predicted) the results. In short, Evo-art is even more open-ended than other types of CG-art.
Interactive computer art (CI-art), in which the form/content of some CG-artwork is significantly affected by the behavior of the audience (see chapter 2), is open-ended too. That is not just because of the complexity of the programs but also because of the unpredictable behavior of the human beings involved in the interaction. Insofar as the latter reason applies, the open-endedness of CI-art cannot be attributed to the computer. As in other (nonevolutionary) cases of C-art, the programmer will be able to predict the general types of art that can result. But the individual cases may be even less foreseeable than they are when they are generated by a computer program running alone, such as AARON.
Most CG-art is open-ended in another sense too, which is that there is no end point foreseen at the beginning of the art-making process. In AARON, for instance, there is no goal state aimed at by the program itself when it starts on a drawing. On the contrary, AARON’s decisions are taken on the fly, sometimes at random and sometimes depending on what other decisions have been made previously.
Open-endedness, in this sense, is not only a criterion of art for Collingwood but a crucial factor in his distinction between art and craft. On his view, craft skills are aimed at some preconceived end. C-art programming does not always fit that constraint. Were he alive today, Collingwood would have to admit that the programming involved in computer art is a highly skilled activity. However, and despite much C-art programming being open-ended, he would probably call it craft, not, as Edmonds does in chapter 10, art. My own view is that programming is not a craft, because crafts exploit universal human dispositions to interpret certain stimuli in ecologically relevant ways (Boden 2000).
Moreover, he would probably point out that this skill is utterly foreign to most human beings. Even people who use computers every day may be wholly ignorant of programming, relying instead on custom-made packages whose interfaces hide all the programming details. So in contrast to his remarks about poets making poets of us all, there is a “distinction of kind” between computer artist and audience.
More accurately, there is a distinction between CG-artists and their audience. By contrast, CA-artists, who rely, for example, on graphics packages such as Photoshop or Brushes, are typically just as ignorant of programming as anyone else. There are exceptions; section 6.4 describes a CA-art system written by AI professionals. For instance, the hugely successful painter David Hockney praised a faster version of Photoshop as “a fantastic medium” (Hockney 2009) and in 2012 held an acclaimed one-man exhibition comprising mostly CA work at London’s Royal Academy of Art. But he draws on the computer screen with a light pen or with his fingers on a touchscreen: no programming is involved.
Even those few CG-artists who get someone else to write their programs for them develop a sense of what is achievable by programming that is not available to the man on the Clapham omnibus. Whereas painting, sculpture, music, and literature all depend on skills that exploit universal psychological properties employed by us every day, programming relies on a highly specialist variety of formal reasoning. This activity is potentially open to everybody, but it remains largely undeveloped in many societies and social groups. Collingwood would probably say that computer art will never be able to touch its audience directly as a result.
As well as granting the skillfulness of computer artists, he would presumably concede that some C-art is decorative, amusing, entertaining, and perhaps even, as demanded by his earlier theory of aesthetics, beautiful. And he might also allow (discussed in section 6.6) that C-art can sometimes seem to express emotions and so can be used in emotionally manipulative ways. He would insist, however, that art proper it is not. But why? And would his strictures apply equally to all forms of computer art?
6.6 Particularism and the Expression of Emotion
Let us consider CA-art first. Human artists, such as Hockney, can use Photoshop as an electronic palette or drawing aid or rely on their own digital photographs or or public domain images on the Internet as a source of items for a collage. There is no reason the results could not express certain kinds of emotion. They might communicate or elicit joy, fear, or sadness just as well as many paintings do. And they might therefore be useful for manipulating people’s emotions in entertainment, advertising, and propaganda. They might even be used to encourage the unexamined “corruption of consciousness” that is involved in popularist politics, such as fascism. It is an interesting historical irony that Max Bense, the first philosopher to write on the aesthetics of computer art and the instigator of the seminal Cybernetic Serendipity exhibition (Reichardt 2008), favored this art precisely because the Nazis—whom he had publicly opposed in prewar Germany—had abused the populist power of emotion in art.
Indeed, one CA-system was developed specifically to depict emotions—namely, anger, disgust, fear, happiness, sadness, and surprise (Colton, Valstar, and Pantic 2008; Colton 2012). This is The Painting Fool, which won the award for the best demonstration at the British Computer Society’s AI conference in 2008. As remarked in chapter 2, the distinction between CA-art and CG-art is not always clear-cut—and the Fool is a CA-system with a generous dose of CG-processes in it. Indeed, its authors see it rather as CG-art, describing it as a “creative collaborator in [the] art project,” as opposed to a system that “merely enhances the efficiency/creativity of [human] users.” They even hope that it “will eventually be accepted as a creative artist in its own right” (Colton, Valstar, and Pantic 2008, 304).
The Fool generates “emotional” portraits by first interpreting the facial expressions of a human being (in person or from a brief video or photograph) in terms of the emotions just listed. This feat requires a complex AI-vision program in itself. Then, having identified the point at which, and the specific facial features by which, the predominant emotion is most strongly expressed, it adjusts the digitized image in numerous ways so as to exaggerate the emotion concerned.
These adjustments are carried out by software designed for rendering visual images in a painterly (nonphotorealistic) way. They include expressive distortions generated by the Fool’s underlying model of twenty-three facial muscles. They also include painting style. For instance, the Fool may choose to depict rage by harsh brushstrokes (in 2-D, not 3-D) of “angry” reds, greens, and black or sadness by gentle strokes with muted colors. And the portrait will simulate some familiar artistic medium—such as acrylic, oils, pastels, water colors, charcoal, or chalk. The brushstrokes characteristic of these different media are defined in terms of seventeen parameters, representing (for instance) brush size, tapering, transparency, friction, stroke length, and varying colors and shades. And the (nonhuman) straight lines generated by the program’s segmentation of the original image are converted into slightly wobbly lines or curves.
The program’s status as CA-art or CG-art is ambiguous because it can be used in different ways. The overall style of rendering may be selected automatically from a list of around 150 predefined sets of rendering parameters (50 for pencil drawings alone), each containing numerous detailed constraints. If left to its own devices, the program may choose a style at random, but it can also opt for an emotionally appropriate medium: pastels for depicting melancholy, vivid colors in “slapdash acrylic” for happiness, or low-saturation grayish colors (indicating rottenness) for disgust. Used in that way, as it was at the AI conference, the Fool can be seen as an exercise in fully automated CG-art. Alternatively, the rendering may be chosen by the user, who (if sufficiently expert) may even specify the style in some detail on the fly. In those cases, The Painting Fool is better seen as a tool for CA-art.
Whether the Fool would have won an award for portraiture, as opposed to AI, is highly doubtful. It is certainly an interesting AI achievement. But its simulated paintings are visually crude in many ways (examples are shown at http://
The authors are well aware of that and plan to improve the Fool in various ways (Colton, Valstar, and Pantic 2008, 311). The main point is that this work shows that painterly programs dealing with emotional images, whether facial expressions or collages of emotive images of other kinds (including images with sociopolitical relevance), are possible.
For Collingwood, of course, a program’s ability to express or elicit emotions in general would not be enough to make it art. But even his particularist aesthetic could, at least in principle, be satisfied by CA-art. This would require that the human CA-artist be able to exploit and supplement the (rich) computer resources with such skill and sensitivity of judgment that the result expressed a particular emotional experience, as may be done by the masterful use of paintbrush or chisel. Hockney’s enthusiastic remarks about Photoshop imply that he thinks this is possible. In this process of conscious construction, the CA-artist would gradually become aware of just what experience it is that is in question, as happens—according to Collingwood—in the noncomputerized case.
Collingwood might object, here, that the emotional expressiveness of the computerized medium is much weaker than that of paint on canvas, for example. He was insistent that art produced by “a man at work before his easel” depended crucially on “the artist’s psycho-physical activity of painting; … his felt gestures as he manipulates his brush, the seen shapes of paint patches that these gestures leave on his canvas” (1938, 307). Put that way, using a light pen on a computer screen, as Hockney does, is perhaps excluded. Painting, he said, “can never be a visual art: A man paints with his hands, not with his eyes. The Impressionist doctrine that what one paints is light was a pedantry which failed to destroy the painters it enslaved only because they remained painters in spite of the doctrine: men of their hands, men who did their work with fingers and wrist and arm, and even (as they walked about the studio) with their legs and toes” (1938, 144–145). Indeed, he saw art in general as based in “an original language of total bodily gesture” (1938, 246).
With no paint in evidence and no muscular activity involved in the production, could a computer-produced image match even “pedantic” Impressionist paintings for expressiveness?
Some CG-art does involve the automated application of liquid paint to physical surfaces. One version of the AARON program, for instance, mixes its own paints and then applies them, with varying quasimuscular pressures, by using half a dozen painting pads of different sizes. It must be admitted, however, that the results do not have the wide range of physical textures seen in human-painted pictures. They are thus much less expressive.
In principle, the CA-artist might be able to do better than this without using actual paint. He or she could turn to 3DP-art (see the taxonomy of chapter 2). That is, he or she could apply the recent techniques of 3-D printing—developed for rapid prototyping in commercial design but now being used by artists too (Sequin 2005)—so as to mimic differing types of brushstrokes on the images printed out by the computer. Although the reflectance properties of paint would not be matched, the 3-D contours of the painted surface would. Each printed layer can be only 0.01 millimeter thick. Those 3-D textures, in turn, could form part of the expressive powers of the medium. A delicately executed Pointillist painting, after all, is less apt for expressing anger, joy, or grief than one painted with the speed and energy, and the generous use of oils, of a Van Gogh. Similar expressive possibilities would presumably inhere in the computer-produced textures concerned. That is especially true for art audiences already culturally familiar with the effects of wielding paintbrushes in different ways and under the influence of different emotions.
That technological strategy (faking brushstrokes of certain kinds where none really exist) might be criticized as betraying a lack of authenticity and so disbarring the result from being a work of art. The concepts of art and authenticity are closely linked (Boden 2007a). In response to such criticism, the human CA-artist could simply drop the faking strategy and rely only on Photoshop or on some other CA-application (such as the Fool) that produces flat images. And to rebut any aesthetic complaints about the flatness—that is, the lack of 3-D texture—of the results, he or she could cite the ColorField painters (e.g., Kenneth Noland), who followed Clement Greenberg’s insistence that the paint should be as flat as possible. Or the artist could say “This is not a painting, but a print. And who is to say that prints cannot be works of art?” even adding “especially one-off, unique prints, like this one.” This assumes that the computer artist is not a follower of Benjamin.
Even if Collingwood were persuaded that what is called C-art is both expressive (of general classes of emotion) and open-ended and even if his mentalist ontology of works of art were set aside (there being no conscious construction in the computer), he still would not accept most C-art as art. The reason would lie in the other key aspect of his theory of art—namely, his particularism.
Why would he reject only “most” C-art? Well, we have seen that CA-art could, conceivably, be particularist. One may doubt whether computer resources will ever exist that are rich enough and computer-using skills strong enough to make this happen in practice. But because CA-art is executed under the close and continuous direction of the individual human artist, there is room—in principle—for a particular emotional experience to be formed, sharpened, and expressed by working within this medium, much as Collingwood described.
But the case is very different for any type of generative art, or G-art. This is defined in chapter 2 as art in which the artwork is generated, at least in part, by some process that is not under the artist’s direct control. As we see there, this category includes noncomputerized examples as different as Mozart’s dice music, Hubert Duprat’s caddis-built jewelry, and Gustav Metzger’s autodestructive bags of rotting rubbish.
The computerized types of G-art, or CG-art, are those in which the artwork results from some computer program being left to run by itself, with minimal or zero interference from a human being. Despite the admitted vagueness of the term “minimal,” this clearly excludes the careful adjustments and sensitive judgments that would be needed to produce something expressing a highly particular emotional experience.
Even in Evo-art, in which the natural selection is typically done by the human artist, the randomness of the rule changes effected by the genetic algorithms would constantly threaten to undermine whatever level of particularist appropriateness had previously been reached. A successfully particularist expression of emotion could occur only as an extraordinary coincidence, and the Evo-program would have to be instantly aborted to prevent its being destroyed.
One could, in principle, cheat. One might first evolve a successfully particularist work of CA-art and then freeze the system so that it could produce the same result automatically. That would fit the letter of the definition of CG-art. But it would not fit its spirit. It would be like waving a magic wand over Leonardo’s brain while he was working on the Mona Lisa so that he then went on to paint it in exactly the same way, over and over again. The real work of particularist emotional expression would have been done only on the first occasion.
More importantly, the particularist intent of the artist as envisaged by Collingwood goes against the grain of CG-art as such. This activity exploits, even exhilarates in, the huge generative potential of the digital computer. That is what “celebrating the medium” means in this context (see section 6.4). A large part of the point of CG-art is to explore, and to exhibit, a certain range of possibilities. Each CG-artwork is unique, yes. But they are unified—both perceptibly and historically—by the program that generated them. The audience is expected to notice that unity and to appreciate it. Whether most audiences actually can notice or appreciate it is discussed in chapter 4.
In the terms used by Davies (2004, chaps. 2–3), GA-works do not “speak for themselves.” Like all art, to be appreciated they must be viewed in a particular art-historical context. Specifically, one in which they have been produced by a computer, not a human hand, and in which indefinitely many other distinct-but-similar works could have been generated too.
In cases in which the CG-artist was aiming to arouse emotions in the audience’s minds, that potential generative unity would be as important as ever. That is, each individually unique emotion-eliciting artwork would be interpreted as an exemplar lying within the CG-range, not as a stand-alone, nicely constructed expression of a highly particular emotion in the human artist’s mind.
A CG-system could generate a range of works eliciting varying experiences lying within one broad type of emotion (fear, for instance). And there already are some CG-programs that can express or elicit instances of a number of different emotions (fear, joy, anger …). Some music-composing programs do this, up to a point (e.g., Riecken 1992). And as we have seen with respect to The Painting Fool, visual C-artists occasionally do so too.
However, recognizing a computer-generated portrait as representing Joe Blow experiencing a moment of fear or joy is not at all the same thing as communicating—or eliciting—a highly specific emotional experience in Collingwood’s (particularist) sense. And that is not the aim of the exercise: insofar as a work by The Painting Fool is considered CG-art, its whole point is that any human face, and very many different emotions, can be represented by it. Pure CA-art is different: Photoshop can be applied to any visual image, but the execution is achieved predominantly by the decisions of the human user. For a CG-artist to write a program capable of generating an artwork eliciting just one, highly individuated, emotion would be a contradiction in terms. Particularism simply is not what is wanted here.
In brief: G-art in general and CG-art in particular may be able to express, communicate, and elicit types of emotion. But it could never construct or express particular emotions, as envisaged by Collingwood. If true art demands particularism, then CG-art is not really art.
6.7 So What?
Collingwood opened his book by saying that he intended to offer a definition of “art,” admitting at once that this term is commonly used equivocally (1938, 1). In other words, his definition would not be a report of common usage but rather what Charles Stevenson (1938) would soon call a persuasive definition. A persuasive definition seeks to direct our attitudes, whether favorable or unfavorable, onto some aspects of the everyday concept concerned while ignoring or underplaying others.
Persuasive definitions are especially likely to be offered with respect to concepts that cannot be defined by a set of necessary and sufficient conditions or by a disjunctive list (so that meeting one or more criteria suffices for something to fall under the concept concerned). That certainly applies to our everyday concept of art. Far from being a neatly definable idea, it is more a matter of overlapping family resemblances (Wittgenstein 1953, 31–33).
What is more, and worse, the concept has what Friedrich Waismann (1951) called open texture, whereby unpredictable extensions are always in principle possible. That is evident in the conceptual changes that have happened in cultural history. For instance, Danto’s definition of art could not have been proffered before the early mid-twentieth century because the art world of dealers, auctions, private galleries, and collectors buying to invest did not then exist. To be sure, Leonardo and Michelangelo depended on wealthy patrons and clients; and the young Picasso was not innocent of the commercial markets of his day. But the art world as we know it now developed only later (Becker 1986).
If all that is correct, some readers may feel that any attempt to define “art” is futile. Collingwood was wasting his time, and we are wasting ours in considering his definition. That is overly dismissive. Even if a cut-and-dried definition is unattainable, a consideration of some of the candidates on offer can help us think clearly about art in general and focus on what is special (and what is not) about computer art.
Collingwood’s account, like all persuasive definitions, respected certain aspects of the everyday notion while downplaying others. I suggest, in this chapter, that his emphasis on open-endedness was overdone: an artist can aim to write a tragedy or a celebratory poem for a wedding. Nevertheless, some degree of open-ended unpredictability is normally thought of as characteristic of art. That is so even though ongoing interaction with the medium can gradually specify a goal of the activity (Harrison 1978).
Some of his other startling claims, too, had close connections with familiar intuitions. Consider his mentalist ontology of “works of art,” for example. Counterintuitive though it may seem, this was intended to emphasize the psychological work done by both artist and audience.
He was not alone in stressing this. For instance, we see in chapter 2 that Marcel Duchamp (whom Collingwood would not have regarded as a proper artist) declared, “The creative act is not performed by the artist alone; the spectator [in deciphering and interpreting it] adds his contribution to the creative act” (1957, 139). In other words, the appreciation of art is almost as creative an activity as its origination and involves many of the same mental processes. It follows that a psychological theory of creativity should ascribe largely similar mechanisms to both artist and audience. For Collingwood, these concerned the conscious conceptualization, by means of language, of preconscious emotions (1938, bk. 2). Alternatively, one may focus on the mental mechanisms underlying creativity (Boden 2004).
In emphasizing the open-endedness of art, Collingwood did not merely mean transformational creativity, which typically causes “impossibilist” surprise (the new structure having been impossible prior to the stylistic change). He also meant the unpredictability that is involved even in the exploration of a familiar style. In addition, and crucially, he meant the unpredictability involved in trying to express a preconscious, largely inchoate, thought or emotion clearly enough to identify it—indeed, to form it—as that thought or emotion.
A further aspect of art that he wanted his readers to favor was its basis in emotion. Put in that way, this too is a commonly accepted criterion, and one that, at first sight, seems to preclude computers from being used as an artistic medium—but see section 6.4. We have probably all heard people refusing to accept Escher’s work as art on the grounds that it is too cerebral, too mathematical—in a word, too unemotional.
But Collingwood’s way of putting it was highly idiosyncratic. That is, this aspect of his persuasive definition—namely, his particularist account of the link between art and emotion—was unusual. Whereas most commentators were content to focus on the expression, communication, and elicitation of general classes of emotion, he was not. For him, art is concerned only with highly specific (and successfully specified) emotions: “Nothing will serve as a substitute. [The artist] does not want a thing of a certain kind, he wants a certain thing” (1938, 17). Moreover, that “certain thing” is an emotion experienced in its full specificity by the artist himself.
Collingwood’s particularism is the aspect of his definition that is least easy to reconcile with computer art. It excludes most CG-art in principle, as we have seen, and practically all cases of CA-art too. However, it also excludes a host of examples that would normally be regarded as art—indeed, as great art. Never mind Escher and Duchamp, both of whom are regarded with deep reservations by many people. Just think of Shakespeare.
Although Shakespeare’s sonnets can be seen as subtle expressions of highly specific emotions in the poet’s mind, his plays cannot. There may be some passages within them in which the poet is expressing particular emotions that he has felt, but in general that is not so.
Consider, for instance, “Uneasy lies the head that wears a crown”—the cry of anguish from Henry IV, tossing restlessly in his perfumed chamber on his sleepless kingly couch. Head of his household Shakespeare may have been, but he never experienced the awesome responsibilities of kingship. Probably, he empathized with the emotions of all his fictional characters, not just Henry. But that is not the same as highly particularized imaginative construction, still less authentic autobiographical experience. Moreover, we have already remarked Collingwood’s insistence that aiming to write a certain kind of text (“a comedy, a tragedy, an elegy, or the like”) is not an artistic activity. Shakespeare’s plays, then, count not as art but as a skillfully crafted combination of entertainment and historical or political propaganda.
In sum, computer art (more strictly CG-art) can satisfy most of Collingwood’s criteria, but it falls down on his particularism. However, if it is excluded as art proper for this reason, then it is in very good company. Whether a persuasive definition of art that has no room for Shakespeare’s dramas is persuasive enough to be accepted is for the reader to decide. The verdict, I suggest, should go against Collingwood.
______________
This chapter is developed from a talk given at the Workshop on Computational Creativity held at the Leibniz Centre for Informatics, Schloss Dagstuhl, Germany, in July 2009.
References
Becker, H. S. 1986. Doing Things Together: Selected Papers. Evanston, IL: Northwestern University Press.
Bell, A. C. H. 1914. Art. London: Chatto and Windus.
Benjamin, W. 1936/1968. “The Work of Art in the Age of Mechanical Reproduction.” Publ. in German in 1936, trans. 1968. Reprinted in Continental Aesthetics: A Reader, edited by R. Kearney and D. Rasmussen, 166–181. Oxford: Blackwell.
Boden, M. A. 2000. “Crafts, Perception, and the Possibilities of the Body.” British Journal of Aesthetics 40:289–301. Reprinted in M. A. Boden, Creativity and Art: Three Roads to Surprise, 50–69. Oxford: Oxford University Press, 2010.
Boden, M. A. 2004. The Creative Mind: Myths and Mechanisms. 2nd ed. London: Routledge.
Boden, M. A. 2006. Mind as Machine: A History of Cognitive Science. 2 vols. Oxford: Clarendon Press.
Boden, M. A. 2007a. “Authenticity and Computer Art.” In “Computational Models of Creativity in the Arts,” edited by J. Jefferies and P. Brown. Special issue, Digital Creativity 18 (1): 3–10. A longer version is in M. A. Boden, Creativity and Art: Three Roads to Surprise, 193–209. Oxford: Oxford University Press, 2010.
Boden, M. A. 2007b. “Creativity and Conceptual Art.” In The Philosophy of Conceptual Art, edited by P. Goldie and E. Schellekens, 216–237. Oxford: Oxford University Press. Reprinted in M. A. Boden, Creativity and Art: Three Roads to Surprise, 70–91. Oxford: Oxford University Press, 2010.
Boden, M. A. 2010. “Personal Signatures in Art.” In Creativity and Art: Three Roads to Surprise, 92–124. Oxford: Oxford University Press.
Carroll, N. 1993. “Essence, Expression, and History: Arthur Danto’s Philosophy of Art.” In Danto and His Critics, edited by M. Rollins, 79–106. Oxford: Blackwell.
Cohen, H. 1995. “The Further Exploits of AARON Painter.” In “Constructions of the Mind: Artificial Intelligence and the Humanities,” edited by S. Franchi and G. Guzeldere. Special issue, Stanford Humanities Review 4 (2): 141–160.
Cohen, H. 2002. “A Million Millennial Medicis.” In Explorations in Art and Technology, edited by L. Candy and E. Edmonds, 91–104. New York: Springer.
Cohen H., and B. Jacobson. 2008. Harold Cohen: Colour Rules (9 July–15 August 2008). London: Bernard Jacobson Gallery, Cork Street. Exhibition catalog.
Cohen, H., M.-E. Lugo, and L. Moura. 2007. AARON’s World: 20 de Junho–28 de Julho 2007. Lisbon: Antonio Prates, Arte Contemporanea. Exhibition catalog.
Collingwood, R. G. 1925. Outlines of a Philosophy of Art. Oxford: Oxford University Press.
Collingwood, R. G. 1938. The Principles of Art. Oxford: Oxford University Press. (References are to the paperback edition of 1958.)
Colton, S. 2012. “The Painting Fool: Stories from Building an Automated Painter.” In Computation and Creativity, edited by J. McCormack and M. d’Inverno, 3–38. Berlin: Springer.
Colton, S., M. Valstar, and M. Pantic. 2008. “Emotionally Aware Automated Portrait Painting.” Proceedings of the 3rd International Conference on Digital Interactive Media in Entertainment and Arts (DIMEA), edited by Sofia Tsekeridou, 304–311. New York: Association for Computing Machinery.
Cope, D. 2001. Virtual Music: Computer Synthesis of Musical Style. Cambridge, MA: MIT Press.
Cope, D. 2006. Computer Models of Musical Creativity. Cambridge, MA: MIT Press.
Danto, A. 1964. “The Artworld.” Journal of Philosophy 61:571–584.
Danto, A. 1981. The Transfiguration of the Commonplace. Cambridge, MA: Harvard University Press.
Davies, D. 2004. Art as Performance. Oxford: Blackwell.
Duchamp, M. 1957. “The Creative Act.” In The Essential Writings of Marcel Duchamp, edited by M. Sanouillet and E. Peterson, 138–140. London: Thames and Hudson, 1975.
Guyer, P. 2003. “History of Modern Aesthetics.” In The Oxford Handbook of Aesthetics, edited by J. Levinson, 25–60. Oxford: Oxford University Press.
Harrison, A. 1978. Making and Thinking: A Study of Intelligent Activities. Hassocks, UK: Harvester Press.
Hockney, D. 2009. “Drawing in a Printing Machine,” at http://www.lalouver.com/exhibition.cfm?tExhibition_id=520. See also Cristina Ruiz, “David Hockney Swaps Oils for Pixels,” Times Online, March 22, 2009, at http://
Kemp, G. 2009. “Collingwood’s Aesthetics.” In Stanford Encyclopedia of Philosophy, edited by E. N. Zalta. Stanford University. Article published March 4, 2009. http://
Lewitt, S. 1967. “Paragraphs on Conceptual Art.” Artforum 5 (10): 79–83. Reprinted in Theories and Documents of Contemporary Art: A Sourcebook of Artists’ Writings, edited by K. Stiles and P. Selz, 822–826. Berkeley: University of California Press, 1996.
Lippard, L. R. 1973. Six Years: The Dematerialization of the Art Object, 1966–1972. New York: Praeger.
Lippard, L. R., and J. Chandler. 1968. “The Dematerialization of Art.” Art International 12 (2): 31–36.
O’Hear, A. 1995. “Art and Technology: An Old Tension.” In Philosophy and Technology, edited by R. Fellows, 143–158. Cambridge: Cambridge University Press.
Reichardt, J. 2008. “In the Beginning …,” In White Heat Cold Logic: British Computer Art 1960–1980, edited by P. Brown, C. Gere, N. Lambert, and C. Mason, 71–81. Cambridge, MA: MIT Press.
Ridley, A. 1998. R. G. Collingwood: A Philosophy of Art. London: Orion Books.
Ridley, A. 2003. “Expression in Art.” In The Oxford Handbook of Aesthetics, edited by J. Levinson, 211–227. Oxford: Oxford University Press.
Riecken, D. 1992. “Wolfgang: A System Using Emoting Potentials to Manage Musical Design.” In Understanding Music with AI: Perspectives on Musical Cognition, edited by M. Balaban, K. Ebcioglu, and O. Laske, 206–236. Cambridge, MA: AAAI/MIT Press.
Sequin, C. H. 2005. “Rapid Prototyping: A 3D Visualization Tool Tasks on Sculpture and Mathematical Forms.” Communications of the ACM 48 (6): 66–73.
Sims, K. 1991. “Artificial Evolution for Computer Graphics.” Computer Graphics 25 (4): 319–328.
Stevenson, C. L. 1938. “Persuasive Definitions.” Mind 47 (187): 331–350.
Todd, S. C., and W. Latham, W. 1992. Evolutionary Art and Computers. London: Academic Press.
Tolstoy, L. 1899. What Is Art? Translated by A. Maude. London: W. Scott.
Waismann, F. 1951. “Verifiability.” In Logic and Language (First Series), edited by A. G. N. Flew, 117–144. Oxford: Blackwell.
Wittgenstein, L. 1953. Philosophical Investigations. Translated by G. E. M. Anscombe. Oxford: Blackwell.