Pluralism, Eliminativism, and the Definition of Art

Traditional monist theories of art fail to account for the diversity of objects that intuitively strike many as belonging to the category art. Some today argue that the solution to this problem requires the adoption of some version of pluralism to account for the diversity of art. We examine one recent attempt, which holds that the correct account of art must recognize the plurality of concepts of art. However, we criticize this account of concept pluralism as being unable to offer an explanation of why some concept is an art concept. Instead, many of the disagree-ments over the definition of art could be reconciled by recognizing that works of art can be valued in a plurality of ways. By recognizing a plurality of values for art, we claim further that the definition of art becomes a non-issue.

The search for art's essence in the philosophy of art has produced no shortage of proposals, and yet no single account has been deemed satisfactory. The general lack of consensus has led to a deep suspicion that an adequate account cannot be reached. The central problem standing in the way of a successful theory is the diversity of objects that strike many as intuitively belonging to art. While many proposals have led to promising insights regarding certain periods, traditions, and forms of art, no single account has yet been able to handle art's plurality. Christy Mag Uidhir and P. D. Magnus have added to the list of proposals with an account that embraces this plurality -on their view, we are mistaken in thinking that there is a single art concept. 1 Rather, they argue, there must be a plurality of art concepts in order to do justice to the diversity of art.
Unfortunately, we find this proposal to be deficient. We hold that their account fails because, for art concept pluralism to be successful, we would need some explanation of what makes any proposed concept count as an art concept. In seeking to account for the plurality of art, their version of art concept pluralism fails to unify the plurality of concepts that they claim belong to art.
We suggest instead that a positive account of art may embrace both value pluralism and eliminativism for the concept art. We hold that the debates that have led to a multitude of proposed accounts of art's essence can be reconciled by accepting a plurality of the ways in which objects can be valued and by recognizing that the concept art does not play a role in picking out some way in which objects can be valued. In this essay, we first review Mag Uidhir and Magnus's proposed account of art concept pluralism. We then offer our objections and consider some possible replies. Finally, we argue that art eliminativism is a plausible alternative to either monist or pluralist definitions of art.

I. From Monism to Pluralism
According to Mag Uidhir and Magnus, the problem with current theories of art is that they all adhere to concept monism. All such theories, including both essentialist and anti-essentialist ones, attempt to rely on a single art concept to account for the wide variety of artefacts that have been deemed artworks. Essentialists, for example, try to capture the nature of art by way of a concept that encodes the necessary and sufficient properties that all instances of art possess. Thus, conventional art defines art as ' artifacts recognized, accepted, targeted, governed by artworld conventions, institutions, and practices' (ACP, p. 91). Anti-essentialist theories also make a bid at capturing the nature of art by appeal to some singular concept; however, they hold that a concept with a more relaxed set of application conditions can account for all artworks. The problem with concept monism is that no single concept can capture the plurality of art. For any proposed singular art concept, regardless of its representational structure (that is, a definition, a prototype, a cluster of examples), Mag Uidhir and Magnus argue that we can always find some artworks that fall outside of it. To continue with the above example, conventional art fails to capture works that are spontaneously created, that reside on the fringes of the artworld, or that are deliberately created with the intention of being anti-institutional.
As a solution, Mag Uidhir and Magnus propose that we adopt art concept pluralism, the idea according to which there are multiple, equally legitimate art concepts that can be employed depending on the context and the task at hand. Their model for such a pluralism is that of species concept pluralism as found in biology, which they consider to be a 'responsible' concept pluralism. According to them, practitioners in the biological sciences employ at least three distinct concepts of species: phenetic species, biological species, and phylogenetic species (ACP, p. 89). All of these species concepts are distinct in that they provide competing accounts of what counts as a species. As such, they address different aims of inquiries and cover different domains of organisms to be studied. For instance, phenetic species appeals to organisms' observable characteristics to divide species, which is beneficial in that it could include every organism within its account. By contrast, if we wanted to divide species by appeal to their reproductive patterns, we would instead employ biological species, which ' distinguishes a species as a reproductively isolated, interbreeding group' (ACP, p. 89). In general, Mag Uidhir and Magnus observe that all of the aforementioned species concepts 'have something in common that makes them all species concepts: we should not expect one fundamental concept to do the work of all the others' (ACP, p. 88). Each species concept has something distinct to offer and that is why we need species concept pluralism.
According to Mag Uidhir and Magnus, the philosophy of art can benefit from adopting such a concept pluralism. Their proposal would be to make room for multiple distinct art concepts, with the expectation that no particular one will do the work of all the others. They recommend the rejection of a singular all-encompassing art concept that aims to account for all instances of art but retains the very idea of an art concept. As they note, 'for concept pluralism to work, there must at least be more than one plausible and productive art concept' (ACP, p. 91). As a start -and this is not intended to be exhaustive -they list four: historical art, conventional art, aesthetic art, and communicative art (ACP,. Each of these art concepts has different application conditions, and thus yields differing extensions (albeit with a great degree of overlap). More importantly, each is useful and productive in distinct ways. For instance, conventional art, which distinguishes art by appeal to artworld conventions and institutions, is useful for inquiries related to sociological and anthropological issues. By contrast, aesthetic art, which identifies artworks as artefacts meeting certain aesthetic considerations, is productive for 'value inquiry and certain cognitive inquiries involving perception' (ACP, p. 92). The important point to keep in mind is that none of these concepts is successful in accounting for all works of art. The key is to 'parcel the work to multiple concepts for the numerous and divergent inquiries pursued by philosophers of art' (ACP, p. 92).

II. Why Art Concept Pluralism Fails
Although Mag Uidhir and Magnus's proposal resolves the worry that no single concept of art can explain all instances of artworks, the central problem with their account is that it is difficult to see how it can make sense of the idea that there can be multiple art concepts. In particular, it fails to provide an account of the conditions under which a particular concept qualifies as an art concept. Without such an account, it is unclear how a pluralism can be generated, that is, how there can be multiple art concepts. To see this, consider an alternative formulation of concept pluralism, endorsed by Daniel Weiskopf, in the philosophy of psychology. It is important at the outset to note that Weiskopf is primarily concerned with concepts as psychological entities or structures, whereas Mag Uidhir and Magnus take concepts to be theoretical abstractions over some domain of inquiry. 2 Despite this difference, we argue that it is instructive to examine Weiskopf's account, specifically on how it makes sense of the idea that there can be a plurality of concepts of the same kind.
According to Weiskopf, current theories of concepts are in a stalemate in that no single theory enjoys wide consensus among psychologists. 3 The problem, in short, is that all of these theories attempt to explain all known cognitive phenomena by appealing to some singular representational structure. Thus, the prototype theory tries to account for such cognitive tasks as conceptual combination, inductive and deductive inferences, and category identification by using a structure that encodes stereotypical properties that its members tend to possess. However, not all of these explanations are supported by the psychological evidence. As Weiskopf points out, there is a wealth of data that indicates that our conceptual system relies on a wide variety of representational structures to perform the relevant cognitive tasks. While prototypes are useful for performing quick identification tasks, people tend to rely on other representational structures to execute other kinds of tasks. For instance, people typically employ causal models to engage in counterfactual and modal reasoning.
To break the stalemate, Weiskopf proposes that we adopt concept pluralism. Instead of insisting that all concepts are reductively identified with some particular representational structure, he advocates that they are constituted by multiple kinds. He offers the following illustration: Having encountered many particular cats, Amira has many cat exemplars stored in memory. As she has encountered particular cats and learned more about cats in general she has constructed a cat prototype that represents the statistical features of cat-kind. At the same time, she believes various things about cats as a natural biological kind, such as that what makes them cats has something to do with what's inside them and what their biological origins are, plus what kinds of transformations they can undergo. Perhaps she regards some of this knowledge as being about what is essential to cats. She also believes that some properties of cats causally depend on others (e.g., purring depends on happiness, meowing depends on hunger). All of this vast body of information coexists in long term memory, but not all of it is activated at a single time. Portions of it, though, are available for ready extraction and use in categorization, building new representations, and guiding actions. When some portion of this information is activated and tokened in working memory, it constitutes one of Amira's cat concepts. 4 The central idea is that, for any category that can be represented in thought, a person can have multiple concepts of it. In the above example, Amira has several cat concepts, each of which is retrieved from long-term memory and accessed to perform the specific cognitive task at hand. Given their diverse functions, none of her cat concepts enjoys a privileged status; they are equally legitimate cat concepts.
What makes all of these representational structures cat concepts? According to Weiskopf, the answer is that they all refer to or represent cats; all of these representational structures have the same extension. What makes them distinct concepts is that each encodes a different body of information, is accessed by different psychological mechanisms, is acquired via different means, and so on. By possessing these distinct representational structures, which each consist of a prototype, a set of exemplars, a causal model, and ideals, Amira has multiple concepts of cats and, thus, multiple ways of thinking about cats. On Weiskopf's view, then, we can clearly make sense of how there can be a plurality of concepts of the same category: all cat concepts are extensionally equivalent, and each concept is distinct in that they are constituted by fundamentally different kinds of representational structures.
Notice that this view of concept pluralism is not open to Mag Uidhir and Magnus's account. The main reason is that the various art concepts as posited by their pluralism have different extensions; the concepts conventional art, aesthetic art, historical art, and communicative art do not refer to the same group of things. By virtue of positing different membership or identification conditions, each refers to a different set of artefacts (albeit with much overlap -we return to this point later). Since these concepts lack a shared reference, they are ipso facto concepts of different things. Therefore, there is no common category from which to generate the pluralism.
Mag Uidhir and Magnus have not provided any reason why we should think of each of these theoretical abstractions as an art concept. Moreover, we have shown that they, unlike Weiskopf, cannot appeal to the sameness of reference or extension as a possible solution. Perhaps there are alternative ways for them to specify when a concept is to be counted as an art concept. Here, Mag Uidhir and Magnus could respond as follows. Recall that their art concept pluralism is based on species concept pluralism. As they say, In the biological sciences, multiple distinct SPECIES concepts are fruitfully employed by practitioners inquiring into various biological matters. Species concept pluralism is the position that these concepts are each legitimate. They have something in common that makes them all SPECIES concepts: we should not expect one fundamental concept to do the work of all the others. (ACP, p. 88) Extending this to art concept pluralism, they could respond that all four of the above art concepts have something in common that makes them all art concepts -namely, we should not expect any of them to do the work of all of the others. This answer, however, is somewhat puzzling, for it does not actually tell us what makes each of them an art concept; that each art (or species) concept has a circumscribed explanatory role has nothing whatsoever to do with why it is a concept of art (or species). At best, an appeal to its limited application might explain why each is unique and why such a pluralism is needed, but this explanation would require one to already have established each as an art (or species) concept. Since this is precisely the issue at question, this is an insufficient response.
Another response that Mag Uidhir and Magnus might offer is that all of the above concepts count as art concepts because they all purport to provide an explanation of what art is. That is, the epistemic goal of each concept is to arrive at an account of art or aims at art inquiries (see ACP, p. 95). Mag Uidhir and Magnus might concede that these concepts share neither intentional content nor representational content, but, since they share the same epistemic goal, they are all art concepts. The problem with this response is that having the epistemic goal of explaining the nature of art is not sufficient to make a concept an art concept. Rather, it makes the concept an attempt to define what art is. Since not all attempts are successful, and Mag Uidhir and Magnus have not provided us with a standard for evaluating whether a concept qualifies as an art concept, these concepts have yet to be demonstrated to be art concepts. Notice that our current demand is not for them to provide us with the concept of art. Instead, what we are asking is that they supply conditions under which a concept would qualify as an art concept. Without such conditions, we would have no reason to think of the above four concepts as art concepts, and, thus, no reason to consider them as constituting a pluralism.
A third response that Mag Uidhir and Magnus could offer is as follows: although conventional art, aesthetic art, historical art, and communicative art have different extensions, there is nevertheless a significant overlap in these extensions. Perhaps these concepts all refer to certain canonical pieces as artworks. As such, they are all art concepts. In our view, this response is antithetical to art concept pluralism, because it seems to suggest that the concept art is constituted by reference to these canonical pieces, with the implication that the qualifiers -conventional, aesthetic, historical, and communicative -are what makes these concepts distinct from one another by adding other artefacts to the canon. In other words, it seems to indicate that the concept art in the unqualified sense is already well understood. But this goes against the principal rationale for advocating pluralism, which is that no single way of understanding the nature of art is sufficient. Consequently, the overlap cannot explain why the above four concepts are art concepts. Another way of establishing this point is to note that some reason must be given as to why the overlap in artworks among these four concepts is significant. Why pay attention to these specific pieces of art? Could they not have resulted from coincidence? It is unclear whether these works are considered to be canonical because they are found in the overlapping space created by Mag Uidhir and Magnus's four candidate concepts, or whether Mag Uidhir and Magnus's four concepts overlap because they are intended to treat these works as canonical. It seems likely that, if some other works had been treated as canonical, other concepts for art would have been more appropriate. Without supplying reasons for giving significance to the artefacts that are at the intersection of the above four concepts, we have no reason to think of them as art concepts.

III. Value Pluralism and the Definition of Art
A plausible account of art (and the art concept), in our view, would do at least two things: (a) it would allow us to make substantive generalizations about (for example) the history and sociology of artistic practice, aesthetic appreciation, and evaluative discourse; and (b) it would do so in such a way that would account for the plurality of those objects covered by these discourses. Competing monistic theories offer various ways of accounting for some generalizations that (a) is concerned with -for instance, expressionist theories offer a way of making generalizations about communication, aesthetic theories offer a way of making generalizations about how objects are evaluated and experienced, and institutional theories offer a way of making generalizations about the sociological practices of the artworld -but, because each of these theories has been developed specifically to make sense of some generalizations and not others, all monistic theories fail at (b). We take it that Mag Uidhir and Magnus's pluralistic approach to the art concept is intended to satisfy both (a) and (b), and, while we reject Mag Uidhir and Magnus's account of art concept pluralism, we accept that the satisfaction of (a) and (b) must demand an appeal to some kind of pluralism. However, we hold that it is not concept pluralism that will save the day. We propose that both (a) and (b) can be satisfied by recognizing that there are a plurality of interests that we take in objects -or a plurality of 'ways of valuing' objects -and that each way of valuing can be picked out by a distinct concept.
To recap, the central issue troubling Mag Uidhir and Magnus's art concept pluralism is that the various concepts they offer as art concepts do not refer to the same extension of things. We accept that each of the proposed individual concepts they offer (for example, historical art, conventional art, aesthetic art, and communicative art) have a substantial role to play with regard to (a). However, we reject their claim that each of these proposed concepts is an art concept (despite their names). Rather, we claim that each of the proposed concepts that Mag Uidhir and Magnus offer picks out a distinct way in which one might take an interest in some object. Take, for instance, the conventional art concept: we hold that this is a concept that picks out a certain interest that one may take with regard to some objects -namely, a sociological or anthropological interest. One uses this concept in order to make certain generalizations about the role that certain objects may play in some society. By contrast, the aesthetic art concept is an entirely distinct concept. It has a different intension, it picks out a different extension of objects, and its use allows a thinker to make different kinds of generalizations about the objects that can be classified under it -namely, generalizations about one's aesthetic engagement with objects. As such, we hold that the aesthetic art concept picks out another distinct way in which one might take an interest in some object.
With this in mind, we believe that we are better able to make sense of the nature of the disagreement between theorists of art. Monist definitions of art are typically proposed on the basis of their perceived superiority in identifying some common feature of those objects that are intuitively thought to be art, and they are rejected when it is discovered that the proposed theory either excludes objects that are intuitively thought to be art or includes objects that are intuitively thought to be non-art. For instance, the conventional art concept picks out certain features of artistic practice and holds that those features are necessary for something to be art, while opponents would reject this account as being insufficient because it fails to account for some of the things that (for example) the aesthetic art concept is able to account for. However, we see this choice between these concepts as illusory -we need not choose; rather, we can simply accept that they are both genuine, but distinct, concepts that play a substantive role in making some, but not other, generalizations. However, and in contrast to Mag Uidhir and Magnus, there is no value in linking these distinct concepts together into a pluralistic art concept. Instead, we suspect that the search for an art concept (as well as the definition of art) could be abandoned and replaced with an inquiry into the many ways in which one might take an interest in some object. What we are recommending, therefore, is a twofold eliminativism: not only should the general concept of art be eliminated, but so should any art concept. 5 It matters not at all whether some object is or is not art. Rather, what matters is what kinds of interests we take in some objects, and how the interest that we take in those objects allows for comparisons to be made to other similarly valued objects.
Consider what it would take to motivate an abandonment of one essentialist theory of art for some other essentialist theory. Imagine a debate between two art theorists, one an imitation theorist and the other an expressionist theorist. The imitation theorist believes that art's function is to imitate the look of real objects through the skilful manipulation of some medium. However, this imitation theorist must also accept that most music would fail to be art as it typically is not an imitation of anything. To the expression theorist, this all seems too narrow. The expression theorist holds that both imitative works and works of music are expressions of emotion, and therefore count as art. The problem, however, is this: the imitation theorist has no reason to accept the criticism that their theory is too narrow as the distinction that their theory is predicated on serves its purpose and identifies as art exactly those objects that exhibit the property that the imitation theorist is interested in. The expression theorist may object that some objects that would count as art on the expression theory fail to count as art on the imitation theory, but this should not concern the imitation theorist one bit. On our view, the choice between these two is illegitimate. Each draws a distinction between objects that the defenders of the respective theories find to be valuable, but each theorist is mistaken in believing that their account captures 'the essence' of art. In debates such as these, the interlocutors are simply talking past one another. Instead, we hold that monist theories of art are typically predicated on a distinct and genuine way of valuing objects, each predicated on a preference for the way in which objects could be valued, and it is precisely for this reason that they fail as definitions of art: each is insufficient for the task of defining art because each fails to establish why that particular way of valuing should take precedence over any other.
Certainly, theorists engaged in the debate over the definition of art take themselves to be doing more than merely stating their preference for the ways in which they value objects. Rather, many theorists engaged in this debate take themselves to be arguing about which way of valuing objects one ought to adopt, or whether some ways of valuing are worthy of esteem and attention. However, we hold that there is no sense in which one ought to adopt one way of valuing over another. Instead, there are many equally valid ways of valuing objects and those different ways of valuing will serve different interests and needs. If theorists wish to argue that some way of valuing ought to be adopted over some others, then we would need an argument showing that one set of interests ought to take precedence over other interests that avoids question-begging. Without this, we are simply engaged in a debate over preferences. This should be resisted.
Thus far, we have been focused solely on essentialist definitions of art -those definitions that propose that there is some single property that is essential to art. However, many theorists today argue for anti-essentialist definitions, which claim that there is no single property that is essential to art but, rather, that art can be defined as (for example) a family resemblance, or by identifying a cluster of non-essential properties, or by having the right social or historical relation to other objects or practices. 6 These certainly would count as single art concept theories and purport to be open-ended enough to account for the diversity of art. However, we believe that these theories should be rejected for reasons similar to those that we have given above against essentialist theories. While this is not the place to launch a detailed criticism of anti-essentialist theories, or to review the extensive literature examining these theories, we offer a brief characterization of each kind of theory and argue that there are good reasons not to adopt either.
One of the most widely discussed arguments for a genetic theory of art is Arthur Danto's argument from indiscernibles: some works of art look indistinguishable from some non-art objects, which suggests that what makes something a work of art is not some perceivable property of it. 7 Rather, Danto argues that some object is a work of art only when the object is viewed under some socio-historically embedded art theory -or an ' artworld'. Moreover, the identity of works, their meaning, and whatever artistic and aesthetic properties they might have also depend on their being viewed under some art theory. Danto's argument may seem to demand that we cannot eliminate the concept art as certain ways of valuing an object must surely depend on a theory of art. 8 However, we hold that the implications of Danto's argument can be interpreted in a way that is not inconsistent with our view. Our account is based on two claims, one positive and one negative. Our positive claim is that there is a plurality of ways of valuing objects, while our negative claim is that there is no overarching concept that unifies these ways of valuing. Danto's argument offers one way of valuing objects. Certain values arise when one regards an object as a product intended for presentation and appreciation. Danto would therefore count as works of art things like Warhol's Brillo Box -indeed, it is the example that birthed Danto's account -however, other theories of art, like Nick Zangwill's creative theory, 9 would not count Brillo Box as a work of art. So, should we side with Danto over the art status of Brillo Box or 'brazen it out' with Zangwill? Or, alternatively, do we have to side with anyone at all? Danto describes a certain way of valuing objects, which would include Brillo Box. Zangwill values objects differently, in a way that does not include Brillo Box. According to our account, Danto and Zangwill disagree about their ways of valuing. We can leave ' art' out of it.
Family resemblance theories hold that something is a work of art because it shares a strong resemblance to previously existing works of art. While not all members will share some feature in common, all members will share some feature in common with some other members. 10 Despite its intuitive appeal, the family resemblance theory suffers a damning problem, as noted by Mandelbaum and Davies: some resemblance relations are irrelevant for membership in the category of art. 11 For instance, a football match that takes place on the same day as a performance of Brahms's Second Symphony does not thereby count as a work of art. Similarly, Andy Warhol's Brillo Box shares many common features with actual Brillo boxes, but actual Brillo boxes fail to be works of art. Being a member of some family requires that its members share certain non-arbitrary relations. But, if we could give an account of which resemblances are non-arbitrarily relevant for membership in the category of art, then the family resemblance theory would no longer be an anti-essentialist theory.
Finally, cluster theories of art seek to identify features that are common to works of art (for example, expressing emotion, being the product of fine skill, being open to interpretation), while claiming that none of those features is necessary for being a work of art. To take one prominent cluster account, Berys Gaut offers ten criteria: (1) possessing positive aesthetic qualities, (2) expressing emotion, (3) being intellectually challenging, (4) exhibiting formal complexity and coherence, (5) having a capacity to convey complex meanings, (6) exhibiting an individual point of view, (7) being an exercise of creative imagination, (8) being a product of a high degree of skill, (9) belonging to an established artform, and (10) having been created with the intention of being a work of art. 12 On Gaut's view, any object that satisfies all of those criteria would unambiguously be a work of art, any object that satisfies none of those criteria would unambiguously fail to be a work of art, and any object that satisfies some of those criteria would be a borderline case.
However, cluster accounts suffer from a problem similar to Mag Uidhir and Magnus's identification of art concepts: either the set of criteria are arbitrary or one must offer a circular defence of that set. One could ask: why those criteria? 13 What makes one set of criteria the correct set? We need some independent reason to view some property as being a relevant property for the cluster. One might argue in favour of some set of criteria on the grounds that the set is associated with paradigmatic works of art, 14 but, clearly, for this to work one must have some independent reason for choosing those works to be paradigmatic examples of art before one identifies the set of criteria associated with them. Thus, we reject the general form of the cluster account on the grounds that no clustered set of criteria can be both nonarbitrary and non-circular.

IV. Art Eliminativism
There are many ways of valuing objects, objects can be compared for the ways in which they are similarly valued, and the totality of these ways of valuing does not form a non-arbitrary class of objects. If this is true, then what does this mean for the art concept? Our view is that, if there can be no monist account of art that can satisfy both (a) and (b) from above (whether essentialist or anti-essentialist), and there can be no pluralistic account of art that offers a way to identify which concepts count as art concepts non-arbitrarily, then we see no further need to maintain that the art concept plays any substantial role. The art concept can be abandoned, and this abandonment would have no impact on discussions of the ways in which objects can be valued. To resist this conclusion, one would need to demonstrate that the art concept in fact does play some substantial role that is not captured by the distinct concepts of ways of valuing. We are highly doubtful that this would be successful. In what follows, we briefly compare our account to Dominic Lopes's 'buck passing' theory in Beyond Art and address some general challenges for eliminativism raised by Annelies Monseré. 15 Our account has much in common with Lopes's theory; however, the difference between our account and Lopes's is significant. Lopes argues that concern over theories of art is largely driven by the 'hard cases' -those objects that force one to ask whether it is a 'work of art' or not -however, theories of art lack the resources to handle such cases. The 'hard cases' tend to function as intuition pumps, but the problem, according to Lopes, is that philosophers' intuitions about the hard cases are theory-laden. Moreover, the methodological criteria one might use to evaluate the strength of different candidate theories are also driven by the same sorts of theory-laden intuitions. As such, the debate over competing theories of art is at an impasse (BA, pp. 53-57). To avoid this, Lopes argues that the hard cases must be explained, by appealing not to a theory of art but rather to theories of the arts. While we may be unable to define ' art', Lopes claims that it is easier to define the individual arts -music, dance, literature, painting, sculpture. An object is a work of art, on Lopes's account, just by virtue of it being a work of one of the arts (BA, pp. 58-61).
We largely agree with Lopes's account of the methodological issues regarding this debate. However, an issue facing Lopes's account is determining the boundary of 'the arts', and it is Lopes's answer to this issue where we break with his account. 16 Music, dance, poetry, painting, and sculpture are surely among the arts, but what about tattooing, gardening, or synchronized swimming? Given Lopes's move to deal with the hard cases by passing the buck onto theories of the arts, this is a question that his account must answer; however, the need to address this question is not unique to Lopes's theory. Wollheim raised what he called the 'bricoleur problem', which is the need to explain why ' certain apparently arbitrarily identified stuffs or processes should be the vehicles of art'. 17 Broadly, there are two strategies to deal with the bricoleur problem: one can either advance a thesis that the arts are unified by some identifiable feature or set of features or one can simply accept the arbitrariness of the collection of practices that are commonly called ' art'. In response, Lopes argues that the arts are a member of a broad category that he calls ' appreciative kinds', which includes those practices that are traditionally identified as the arts, along with things like gardening, high diving, and clowning. What unites the appreciative kinds together is that they involve evaluative practices where objects can be judged as better or worse examples of their kind. As Lopes says, an ' art, K, is an appreciative kind just in case there is a property of being good qua K, or being good-modified for a K, or being good qua K* for a K' (BA,p. 146). Landscape painting is one appreciative kind. There are certain properties one can look for to evaluate whether a particular object is good qua the kind landscape painting. Gardening is another appreciative kind. There are other properties one can look for to evaluate an object qua the kind garden. So, is gardening an art-kind? Lopes considers two answers to this: to determine whether some appreciative kind is an art-kind, (1) we may rely on analogies and disanalogies that it holds to other art-kinds (BA, pp. 115-18), but (2) such a process may only demonstrate that 'the arts' is in fact an arbitrary grouping of appreciative kinds (BA,.
However, neither of these answers will work. According to the first, we can determine whether appreciative kinds are art-kinds by looking for analogies and disanalogies between them and the established arts. For instance, film can be considered among the art-kinds because it holds certain strong analogies to theatre, which is surely an art-kind (see BA,. The problem with this approach is that one can only draw an analogy to theatre if we have already accepted theatre as an art-kind. But we need an argument for why theatre should be among the art-kinds in the first place. We cannot appeal to analogies or disanalogies between the various kinds without knowing what makes the established art-kinds art. Lopes could then fall back on his second proposal, which accepts the arbitrariness of 'the arts'. 'Appreciative kinds' is a very wide class -it includes the traditional art forms as well as 'natural objects and settings, the crafts, industrial design, and much else besides' (BA, p. 124). On this proposal, there is no further need to distinguish between art-kinds and non-art-kinds. The problem, however, is that passing the buck onto the arts offers no way to account for the hard cases. Is a particularly stunning piece of topiary a work of art? If 'the arts' is an arbitrary class, then one could only say that topiary is one of the appreciative kinds while admitting that there is no distinction between art-kinds and non-art-kinds (BA,p. 124). But this leaves the 'is it art' question unresolved. 18 One of the main benefits of the buck passing theory is supposed to be its ability to deal with the hard cases, like topiary, in which case we should not accept 'it's arbitrary' as an answer to the bricoleur problem. Lopes's account is therefore unable to deliver on its central promise.
Finally, Monseré offers a general challenge to any eliminativist account. According to Monseré, art is both a structural and normative concept, and therefore we cannot do without it. Art is a structural concept in the sense that it is central to many different kinds of concerns. Academic fields (like cultural studies), institutions (like art museums), and intellectual pursuits (like art education) all make some place for the concept art. As Monseré notes, it would be difficult to imagine these fields without employing the concept. 19 Additionally, art has certain normative functions. As Monseré says, ' categorizing an artefact as art brings with it a cluster of normative attitudes towards it, regarding its preservation, treatment, and evaluation'. 20 For instance, if comic books should count as art, then we should not quickly dismiss them as childish entertainment. Importantly, there are cultural consequences to accepting or denying an object's status as a work of art, and the works of underrepresented and oppressed groups are often denied this status. If we eliminate the concept, then Monseré worries that we would be left with no principled way to right these wrongs. This is unacceptable as 'the arbitrariness of the grouping of the arts is by no means innocent; indeed, it has serious implications for the treatment and evaluation of practices of minority groups'. 21 We will raise three points in reply.
First, the central point that motivates our art eliminativism is the observation that different candidate art concepts pick out different extensions of objects. While it may superficially appear that art is a structural concept linking diverse areas like cultural studies, art education, and art practice, a closer look almost always reveals that the scholars and practitioners who operate within these areas are concerned with different classes of objects. Indeed, many of those scholars reject the relevance of art and prefer terms like ' cultural studies' and 'media studies' for precisely this reason. Given this, it would be more productive for scholars and practitioners to define the extension of objects that interests them without worrying about having to carry the burdens of art.
Second, we wish to clarify Monseré's suggestion that art carries certain normative functions. We think it would be more accurate to say that there is a diverse range of non-overlapping concepts that each carry their own distinctive normative functions. For illustration, consider two of Monseré's claims. Monseré draws attention to the 'normative attitudes' that concern art's 'preservation, treatment, and evaluation', 22 while, earlier in her essay, Monseré also claims that, '[while] not all art is good art, what it means to be art is strongly connected to what it means to be good art'. 23 Notice how different these normative roles are. The kind of normative attitudes that regard art's preservation and treatment are the same sort of attitudes that would attach to non-art cultural objects as well. In which case, talk about the appropriate means of preservation, treatment, and evaluation of some object likely falls under one sort of normative concept -call it the cultural significance concept -while talk about what makes something 'good art' likely falls under another sort of concept -perhaps a concept like aesthetic pleasure. When viewed this way, our claim is that there can be no overarching 18  concept art that captures the extension of both of these normative functions. Instead, these are distinct concepts with non-overlapping extensions that do their own normative work.
Finally, we entirely agree with Monseré that we should worry about the cultural uses and abuses of art. However, we believe that an eliminativist account can do the job. The concept of art that we have inherited from the intellectual tradition following Charles Batteux is one that was defined by the wealthy elites and colonial powers of Europe. 24 That concept was in part used to justify the colonization of non-European peoples on the grounds either that their cultures did not produce art or that the art they did produce was substandard. The failure of some cultures to meet elite European standards of art was evidence that those cultures were backward and in need of a civilizing hand. There seem to us two ways to counter this problem: either find a non-arbitrary grouping for art broad enough to accord the same reverence and respect for all cultures' works, or eliminate this colonialist concept. Art eliminativism does not reproduce old biases but rather offers a way to transcend them.
In closing, our proposal is that there is a plurality of substantive generalizations that one can make about the ways in which objects can be valued, and that the recognition of this point would eliminate the art concept as well as the need to define art. If our proposal were adopted, we see a number of advantages. First, our proposal could easily satisfy both (a) and (b) from above. An account of the ways in which objects are valued would allow us to make substantive generalizations about objects by allowing us the space to classify objects according to the ways in which they are valued, thus satisfying (a), and our proposal would account for the plurality of objects that are valued by recognizing that these ways of valuing do not make up one non-arbitrary class (or a non-arbitrary cluster) of objects, thus satisfying (b). We accept that there may be much overlap between the ways in which objects are valued; however, we would insist that valuing an object in one way would not necessarily lead to, or preclude, that object being valued in some other way. We expect that there would be much overlap between those objects that are valued for their ability to afford one an aesthetic experience and those objects that are valued for their ability to communicate some content (comp. ACP, pp. 91-92), but valuing some object aesthetically should not thereby suggest that that object would (or would not) be valued communicatively.
Second, our proposal would make some debates within the philosophy of art obsolete. For instance, looking through the literature today, one will find an overabundance of essays arguing that some group of objects is (or is not) art. The art status of comic books, 25 fashion, 26 food, 27 gardens, 28 perfumery, 29 pornography, 30 screenplays, 31 and video games 32 have been recent topics of debate, and one could easily imagine this debate extending to interior decorating, cosmetics, plastic surgery, tattooing, and body piercing. On our view, it is of no value to argue whether some object is art or not. Rather, it is of great value to argue whether some object is, or could profitably be, (for example) appreciated aesthetically, or viewed as part of the canon of some particular history of production or performance, or regarded as an object that has been offered as a candidate for appreciation within some sociological context. These are questions that can be genuinely examined and debated, and, to address these questions, it seems to us to matter not at all whether some object is art or not.