Sunteți pe pagina 1din 16

British Journal of Aesthetics, Vol. 45, No. 3, July 2005 British Society of Aesthetics; all rights reserved.

For Permissions, please email: journals.permissions@oupjournals.org


doi: 10.1093/aesthj/ayi032

THE CLUSTER ACCOUNT OF


ART DEFENDED
Berys Gaut
This paper replies to objections from Thomas Adajian, Stephen Davies, and
Robert Stecker to my claim, defended in Art as a Cluster Concept, that art
is a cluster concept and so cannot be dened. The paper also claries and extends
the arguments of the earlier paper and locates its position in relation to the work
of Morris Weitz.

i. the cluster account

In Art as a Cluster Concept I argued for the view that art is a cluster concept and for that reason cannot be dened.1 This claim has occasioned a range
of critical commentary. Thomas Adajian has argued that the cluster account is
inferior to denitions of art; Stephen Davies and Robert Stecker have held in
contrast that, far from being an alternative to a denition, the cluster account
is an instance of one.2 In the present paper, I reply to these always interesting
and often probing objections, and argue that the cluster account of art is superior to denitions of art. I also clarify some of my previous claims, extend and
develop some of the earlier arguments, and discuss my view in relation to
Morris Weitzs claim that art is an open concept.3
In holding that art is a cluster concept, I mean that there are multiple criteria
for the application of the concept, none of which is a necessary condition for
somethings being art. A criterion is a property, possession of which conceptually
1

Berys Gaut, Art as a Cluster Concept, in N. Carroll (ed.), Theories of Art Today (Madison:
University of Wisconsin Press, 2000), pp. 2544. Henceforth I will refer to this paper as
Cluster in the main text.
Thomas Adajian, On the Cluster Account of Art, British Journal of Aesthetics, vol. 43 (2003),
pp. 379385; Stephen Davies, The Cluster Theory of Art, British Journal of Aesthetics, vol.
44 (2004), pp. 297300; and Robert Stecker, Is it Reasonable to Attempt to Dene Art? in
Carroll (ed.), Theories of Art Today, pp. 4564.
Morris Weitz, The Role of Theory in Aesthetics, Journal of Aesthetics and Art Criticism,
vol. 15 (1956), pp. 2735; reprinted in P. Lamarque and S. H. Olsen (eds), Aesthetics and
the Philosophy of Art: The Analytic Tradition (Oxford: Blackwell, 2004), pp. 1218. Page
references in the rest of this paper are to the reprinted paper.

British Society of Aesthetics 2005

273

274

CLUSTER ACCOUNT OF ART DEFENDED

counts towards an objects falling under a concept; there are several criteria for a
concept. The notion of counting towards I construe as follows. First, if all of the
properties that are criteria are instantiated, this sufces for an object to fall under
the concept; and more strongly, if fewer than all of these properties are instantiated, this also sufces for the application of the concept. So there are jointly
sufcient conditions for the application of the concept. Second, there are no
properties that are individually necessary conditions for the object to fall under
the concept (that is, there is no property that all objects falling under the concept
must possess). Third, there are disjunctively necessary conditions for application
of the concept: some of the properties must be instantiated if the object is to fall
under the concept. By the second point, it follows that if a concepts meaning
is given by a cluster account, one cannot dene that concept, in the sense of
xing individually necessary and jointly sufcient conditions for it.
I defend a particular instance of the cluster account in its application to art.
This involves ten criteria that count towards an objects being art: (i) possessing
positive aesthetic qualities (I employ the notion of positive aesthetic qualities
here in a narrow sense, comprising beauty and its subspecies); (ii) being expressive
of emotion; (iii) being intellectually challenging; (iv) being formally complex
and coherent; (v) having a capacity to convey complex meanings; (vi) exhibiting
an individual point of view; (vii) being an exercise of creative imagination;
(viii) being an artefact or performance that is the product of a high degree of
skill; (ix) belonging to an established artistic form; and (x) being the product
of an intention to make a work of art.4
One thread in Adajians objections is that the cluster account is in various
ways no better than some denitions.5 I will shortly query that. But even if it
were true, something important would still have been established. For the
standard story of attempts to dene art in the recent analytic tradition starts
with a discussion of Weitzs seminal paper, argues for the failure of his appeal
to the notion of family resemblance and of his claim that art cannot be
dened, and then moves swiftly on to discuss various denitions, such as aesthetic, institutional and historical ones. The failure of a broadly Wittgensteinian
position, which holds that art cannot be dened, is thus taken as a given in
the subsequent quest for denitions. Admittedly, there have been dissenters to
this story: Stephen Davies, with his characteristic erudition in these matters,
notes that there have been intermittent attempts to defend a cluster theory.
But it is nevertheless true that the consensus holds that the Wittgensteinian
4

Art as a Cluster Concept, pp. 2628. I also hold that there is one necessary condition for
somethings being an artwork, namely its being the product of an action. This applies by virtue of the notion of a work, rather than of art. So, strictly, the view defended is a modied
version of a cluster account.
Adajian, On the Cluster Account of Art, p. 385.

BERYS GAUT

275

project failed, a consensus that Daviess own inuential book about the denition of art endorses.6 So even if the cluster account were not shown to be
superior to denitions, but merely shown to be viable and not to succumb to
standard objections, something signicant would have been established, for
the standard story would thereby be called into question: we should not simply
take it as a given that anti-denitionalism is a failure. We should retell the
standard story so that a door opened by Weitz is not assumed to be closed.
A fundamental move is to distinguish two forms of the family resemblance
view, which are often conated. One version is the resemblance-to-paradigm
account: this holds that something is art by virtue of resembling paradigm artworks. The second versionthe cluster accountholds that something is art
by virtue of satisfying a range of criteria. The rst version falls to standard
objections, such as that the account is vacuous (since anything can resemble
anything else in some way) and that it is incomplete (since we have not been
told what the paradigms are). The second version avoids these objections: by
stating what the criteria are, vacuity is avoided, and by not appealing to paradigms, incompleteness does not threaten the account. None of the critics of
the account has queried this distinction; yet it shows that some of the basic
objections to the family resemblance view can be avoided by one version of it,
and therefore that an important account of the concept of art is still viable.
ii. counterexamples and arguments

I distinguished above between the cluster account per se, and a particular
instance of it: the ten criteria that I propose as a plausible example of the
account. I also noted in Cluster that if a plausible counterexample to the particular version of the account were established, the version could be modied.
Undermining the particular version does not show that the cluster account
per se is mistaken.
Adajian writes that I thereby appear to deny that defective substitution
instances of the cluster-schema count against the cluster-schema. And he
objects that in parallel fashion the defender of a denition of art could say that
defects of particular denitions do not show that a denition per se is impossible.
It follows that I cannot appeal, without self-contradiction, to an argument
from induction to show that denitions of art fail. And that leaves me with
scarce resources for defending an argument for the indenability of art.7
This objection is based on a misunderstanding of what I claimed in Cluster:
my simple point there was that, just because a particular instance of the cluster
6

Davies, The Cluster Theory of Art, p. 298; and Davies, Denitions of Art (Ithaca, NY:
Cornell U.P., 1991), ch. 1.
Adajian, On the Cluster Account of Art, pp. 381382.

276

CLUSTER ACCOUNT OF ART DEFENDED

account is shown to be defective, that does not entail that all versions of the
account are defective. In the same way, just because a particular instance of a
denition (say, an institutional theory) is defective, this does not entail that all
denitions are defective. It is a long way from making this point to holding
that no number of counter-examples to particular cluster accounts (or denitions) could give one reason to doubt the correctness of cluster accounts (or
denitions). It is the latter, stronger claim, not the more modest one, which
would rule out my consistently using an argument from induction against
denitions. But I did not advance the stronger claim in Cluster. And my
point about particular denitions was that the cluster account, in the version I
favour, sidesteps some of the salient difculties of well-known denitions of art.
Still, Adajian raises the question of induction and it deserves an answer. Does
the continued failure to nd an agreed-upon denition give one grounds to
believe that one cannot dene art? It is certainly the case that this failure does
not entail that one cannot dene art. There are other, plausible explanations
possible for our continued failure: perhaps we are not clever enough to nd the
correct denition, or have not looked long enough, or perhaps we have found
the correct denition already, but not everyone has recognized it as such. Does
the failure to nd an agreed-upon denition make it highly probable that art
cannot be dened? Given the wide variety of kinds of denitions that have
been proposed, we cannot be condent that there are not other sorts of denitions, which we have not yet considered; and in comparison to the potentially
innite number of candidate denitions, those that have been examined so far
are nugatory in number. The same points about lack of intelligence, effort, or
recognition apply too. So an inductive argument that concludes with the truth
of the claim that there is no denition of art looks untenable.8
Although an epistemic version of the argument from induction fails, there is
a more modest, heuristic version, concerning reasonable search principles.
Take a simple, homely example. Suppose that I am looking for my car keys. I
have searched the house carefully. Rather than take up the oorboards in
search of them, it would be reasonable to look in some other places rstfor
instance, to see if I have left them in my car. If something for which I am looking could be in one of two places, my continued failure to nd it in one of
them makes it reasonable to look in the other one, if I have searched that second place hardly at all. The parallel to looking for an account of art is clear: if
we have been predominantly looking for a denition of art, but have hardly
examined cluster accounts at all, then continued failure to nd a denition
makes it reasonable to start looking at cluster accounts. We can extend the parallel. Perhaps I have convinced myself that my keys cannot be in the car, since
8

Robert Stecker is similarly sceptical about inductive arguments against denitions in Is It


Reasonable to Attempt to Dene Art?, p. 54.

BERYS GAUT

277

I am sure that I had them in my hands when I entered the house, and that is
why I am searching the house. But then I may realize that this was not so: what
I remember having in my hands were my house keys. Similarly, someone may
have been convinced that family resemblance accounts were untenable, because
of well-known objections: if, as I have argued, the cluster account avoids these
objections, then the reason for not looking at cluster accounts for the correct
account of art are shown to be untenable, and one should look there.
The heuristic version of the induction argument does not conclude with the
truth of the cluster account, but merely that it is reasonable to put more effort
into looking for a version of the cluster account, given the failure to find an
agreed-upon denition and the fact that the cluster account does not succumb
to the standard objections. But as far as the argument goes, the cluster account
might still be false (just as my keys might really be in the house, after all). The
heuristic conclusion is all that one can safely draw from inductive considerations.
What argument can be found for the truth of the cluster account, if not an
inductive one? Weitz developed an argument from creativity: he held that
the very expansive, adventurous character of art, its ever-present changes and
novel creations, makes it logically impossible to ensure any set of dening
properties.9 The fact that art is creative entails that one cannot dene it, since
to dene something is to set limits to it, and the creative artist can breach these
limits. This argument is unsound for several reasons: for instance, it might be
part of the denition of art that it is a creative practice; and practices can be
creative, even though they can be dened (physics, for instance).10
A different argument is available. When Wittgenstein introduced the notion
of family resemblance, he wrote Dont say: There must be something common, or they would not be called games but look and see whether there is
anything common to all.11 Weitz quotes this passage in arguing for the family
resemblance view. Call this appeal to looking and seeing the argument from
inspection. The point is simple: one should not just assume that there must
be a denition of art; rather, one should examine actual and counterfactual
cases, and see whether a family resemblance view captures them correctly.
Mandelbaum in his inuential critique of Weitzs paper took it that look and
see meant that only manifest, non-relational, properties were to be examined.12

9
10
11

12

Weitz, The Role of Theory in Aesthetics, p. 16.


See Davies, Denitions of Art, ch. 1, for a comprehensive rebuttal of Weitzs argument.
Ludwig Wittgenstein, Philosophical Investigations, third edn, trans. G. E. M. Anscombe
(Oxford: Blackwell, 1978), part I, proposition 66.
Maurice Mandelbaum, Family Resemblances and Generalizations Concerning the Arts,
American Philosophical Quarterly, vol. 2 (1965), pp. 219228; reprinted in Alex Neill
and Aaron Ridley (eds), The Philosophy of Art: Readings Ancient and Modern (New York:
McGraw-Hill, 1995), pp. 192201.

278

CLUSTER ACCOUNT OF ART DEFENDED

But nothing in Weitzs, still less in Wittgensteins, account requires the method
to be understood with this restriction. Rather, it is simply the standard philosophical method of thought-experiments.
Why, when one looks and sees, though, should one discover that a cluster
account ts the concept of art? The answer is partly given by factors to be discussed later: cluster accounts can handle borderlines well, they can give good
error theories about the appeal of theories and of conicting intuitions about
disputed cases, and so on. But, importantly, the cluster account also ts our
intuitions so well because it allows us to acknowledge the relevance of the
diversity of grounds that we give in common talk about why something is or
is not an artwork. When we argue about whether something is art, we appeal
to a variety of criteria: remarks such as my child could do that show that skill
is a factor; the distinction between art and entertainment shows that we expect
art to give us something more than pleasure, so that it is informative or intellectually challenging; we point to the creativity of a piece as a ground for
thinking that it is art; we hold that it counts towards somethings being an artwork that it is in an established artistic form, but that pieces falling outside
such established artistic forms (for instance, chairs) can qualify as art if they
exhibit a high degree of qualities such as creativity and beauty; and so on.
Arguments about disputed cases of art similarly reveal general criteria to which
we commonly appeal in establishing whether something is art.13 The sheer
variety of artworks and art forms also suggests that only a complex account,
involving multiple criteria, will easily t the phenomena. So cluster accounts,
through their multiple criteria, allow us to take at face-value much of our
common talk about why something is art and to acknowledge the enormous variety of artworks. Though their philosophical development may be
complex, the underlying motivation of cluster accounts is simple.
Adajians objection, though, was that the view had scarce resources to
show that art is indenable, rather than specically to show that the cluster
account is tenable. And how could one show that art is indenable without
appeal to some very general considerations such as the induction or creativity
arguments? The answer is straightforward: as noted in section I, the truth of
the cluster account entails that art cannot be dened, in the sense of giving
individually necessary and jointly sufcient conditions for somethings being
art. So the argument for the indenability of art follows immediately from the
argument for the cluster account. Nothing more is required.

13

See Art as a Cluster Concept, p. 28. Historical denitions can also allow for a variety of
factors that are relevant to somethings being art and thereby have some of the merits of
cluster theories of art. But they allow for this variety of factors only indirectly and fall to objections about the projectability of our concept of art. On the latter, see Gregory Currie,
Aliens, Too, Analysis, vol. 53 (1993), pp. 116118.

BERYS GAUT

279

iii. borderline cases

I claimed in Cluster that one respect in which the cluster account shows itself
to be linguistically adequate is in the way it explains cases that are borderline to
works of artfor instance, cookery. Adajian objects that no explanation at all
is offered by the account. For the cluster account holds that the cases are
borderline in satisfying some subset of the criteria, but equally, it holds that
some artworks satisfy some subset of criteria, as do some non-artworks. Since
borderline cases of art, clear artworks, and clear non-artworks may all satisfy a
subset of criteria, no explanation is given by my version of the cluster account for
borderline cases. This is in contrast to versions such as Weitzs, which hold
that most of the criteria must be satised to make something art. Moreover,
denitions can handle borderline cases perfectly adequately, since they may involve
vague terms, and borderline cases could fall into the area of indeterminacy.14
Cluster accounts which hold that most of the criteria must be satised in
order for something to be art can indeed provide a method for determining
whether something is a borderline case of art: they can stipulate what proportion of criteria must be satised in borderline cases, and what proportions in
the cases of works that are clearly art and works that are clearly not. But such
accounts are implausible: why should we suppose that our concepts are structured so that each criterion is of equal importance no matter what the other
criteria are with which we consider it? Weitz does not give a list of criteria
(indeed, his view seems to run together the resemblance-to-paradigm and
cluster accounts). But Denis Dutton does: he has defended a list of eight criteria that sufce for something to be art.15 His criteria for an object or performance to be an artwork are, roughly, that (i) it is a source of pleasure in itself;
(ii) it requires the exercise of a specialized skill; (iii) it is made in a recognizable
style; (iv) there exists a critical language to discuss it; (v) it represents in varying degrees of naturalism; (vi) its makers consciously intend it to produce
pleasure; (vii) it is frequently bracketed off from ordinary life, being made
special, often involving intense emotion; and (viii) it affords an imaginative
experience. None of these, holds Dutton, is a necessary condition for an object
or performance to be an artwork, but if most of these criteria are satised by
something, then it is an artwork. For instance, Yoruba twin sculptures are artworks by virtue of satisfying six of these criteria: (ii)(v), (vii), and (viii).
Duttons account has much to commend it, and partly overlaps with the
account defended here. There is a problem, though, in its use of a numerical
14
15

Adajian, On the Cluster Account of Art, pp. 382383.


Denis Dutton, But They Dont Have Our Concept of Art , in Carroll (ed.), Theories of
Art Today, pp. 217238. Dutton thinks that there are other potential candidates for his list.
It is not clear whether he understands his account as a denition or as cluster account, but
this does not matter for the point under discussion.

280

CLUSTER ACCOUNT OF ART DEFENDED

threshold for establishing whether something counts as art: some other sets of
six criteria do not sufce to make something art. For instance, professional
football games satisfy (i)(iv), (vi), and (vii), yet professional football games are
not works of art. So some sets of criteria sufce for somethings being art, but
an equinumerous set with different criteria does not. In the example mentioned, it looks as if representation (condition (v)) is important (and indeed,
the criterion of being an accepted art form that does not appear on the list
seems to be important here too). One could try to get round this problem by
weighting some criteria more heavily than others: but even if one could nd
some satisfactory method for weighting criteria, there is no reason to assume
that criteria have the same weight across all possible subsets. It is for this reason
in part that I cast my version of the cluster account in disjunctive form. The
appeal to the presence of most or weighted criteria is avoided, and one instead
lists disjuncts and determines whether it is true of each disjunct that it sufces
to make something art, makes something borderline art, makes the thing
clearly not art, or is completely indeterminate.
Hence there are reasons not to stipulate a threshold number of criteria
required for something to be art, and so reasons not to do this for borderline
cases. The cluster account, as I have formulated it, has the implication that
objects satisfying all or none of the criteria will not be borderline cases of art;
but that is all. So it is certainly the case that merely saying that an object satises
some but not all of the criteria does not differentiate between art, non-art, and
borderline cases. How, then, does one determine which subsets of criteria make
something a borderline work of art? The method is the familiar method of
inspection: that is, consider the particular subset, and consider whether something satisfying it is a borderline case or not.
If we employ the method of inspection, it may seem that we have lost any
explanatory role for the cluster account in respect of borderlines. But that is
not so. For the cluster account explains why, if something is a borderline case,
it has the particular form it has, and can explain what makes it a hard case.
Consider cookery: cookery seems to be an art in satisfying the criteria, for
instance, of involving the exercise of creativity, having positive aesthetic
properties, and being the product of skill; but it fails in respect of conveying
complex meanings, expressing emotions, and is usually not the product of an
artistic intention. In thinking about whether cookery is an art, these seem to be
just the sort of factors we take into account: we are pulled towards saying that
it is an art by virtue of the rst three criteria, but feel resistance by virtue of the
latter three. The hardness of the case is preserved.
Adajian points out that a denition could handle borderline cases by locating them within the vagueness established by some of the terms in the denition. That is correct, and some borderline cases or aspects of them may indeed
be like that. Since the terms within cluster accounts may also be vague, the

BERYS GAUT

281

cluster account can also capture what we can call vagueness borderlines. For
instance, if aesthetic is vague term, then both an aesthetic denition of art
and a cluster account of the kind recommended (in respect of its rst criterion), could allow for some vagueness in the notion of the aestheticfor
example, as to whether tastes have aesthetic properties. But the cookery case
is not predominantly like that. It is clearly true that cookery can involve a high
degree of skill and it is clearly true that it cannot convey complex meanings.
There is no vagueness about whether these criteria are satised or not in this
case. Here we have a conict: clearly satised or unsatised criteria pull in
different directions. Call this kind of case a conicted borderline. The cluster
account can allow for these kinds of borderlines and explain why they are hard
and have the particular form that they do. Herein lies the extra explanatory
power of the cluster account. Note too that cluster accounts are also committed to holding that any other practices, which satisfy these particular criteria
and fail to satisfy the others, are also borderline cases. Hence generality and
prediction are involved in this classication: if it transpired, for instance, that
some art forms satised the same disjuncts and failed to satisfy the same ones
as did cookery, then one would be rationally required either to reconsider the
borderline classication of cookery, or the classication of the putative art
form as art. Here as elsewhere, the cluster account has normative bite.
iv. normative adequacy and explanations

I claimed in Cluster that any adequate account of a concept must prove itself
to be not only adequate to intuition, but also be normatively adequate: that is,
the account has to be broadly consistent with our linguistic intuitions, and must
also, where it differs from some of those intuitions, explain why they are mistaken (it must provide a theory of error). Since there are disagreements about
whether, for instance, conceptual art, primitive art and popular music are
really art, the account should provide a plausible theory of error about the intuitions that conict with those that it favours. Ideally, it should also explain the
appeal of denitions, and show that they are mistaken. The cluster account can
meet these desiderata. It can explain the appeal of many denitions: they take
a legitimate criterion for somethings being art and inate it into a necessary
and sufcient condition. The account can likewise explain many disagreements
about what counts as art, and resolve them. Those who claim that the disputed
cases are not art may do so because they insist on, as a necessary condition for
art, some feature, such as being the product of an intention to make art. But
it is a mistake to insist on this as a necessary condition, as can be shown by
considering less contentious cases of art, which lack this feature.16
16

Art as a Cluster Concept, p. 3638.

282

CLUSTER ACCOUNT OF ART DEFENDED

Adajian objects that it is not at all clear that these are genuine explanations,
and in any case parallel explanations can be offered by the friend of denitions. Whereas the cluster theorist holds that her opponent mistakenly converts mere criteria into necessary and sufcient conditions of arthood, the
friend of denitions holds that the cluster theorist mistakenly converts such
conditions into mere criteria. The situation is symmetric: there is no explanatory advantage for the cluster theorist.17
These explanations are, however, genuine: they are not vacuous, nor do
they appeal to processes or entities that are incoherent or unknowable. So the
more weighty complaint is that the cluster theorist has no explanatory advantage over the friend of denitions. But that is also not true.
Consider rst the explanation of the appeal of competing denitions. The
cluster account holds that in many denitions a single criterion of arthood is
illegitimately converted into a necessary and sufcient condition: a complex
situation (in this case, the existence of many criteria) is converted into a simple
one (that there is only one condition). The cognitive bias towards simplication is a common one, to be found operating in a variety of prejudices. In
contrast, there is no similarly common bias towards making simple situations
complex, yet that is what the believer in these denitions would have to
appeal to in order to give a parallel explanation of the appeal of the cluster
theory: the conversion of a single condition into multiple criteria.
The cluster account can also explain why a wide range of denitions is
appealing. Many denitions fasten onto a single criterion, and inate it the
essence of art: thus expression theorists latch onto the expression criterion,
aesthetic theorists onto the beauty criterion, formalists onto the formal
criterion. All of these criteria play a role in the cluster account. No similarly
wide-ranging explanation is available to the believer in simple denitions,
since the formalist, for instance, cannot employ the same mechanism to explain
the attractions of a denition of art in terms of expression, for she countenances
no constitutive role for expression in her denition of art.
Consider next explanations concerning disagreements about whether something is art. The cluster theorist offers genuine explanations, which have the
same form as those noted above: the denitionalist has simplied the situation,
latching onto a single criterion and inating it into a necessary condition.
Suppose, for example, that someone holds that art is dened in terms of
makers intentions to create art, and concludes that primitive art is not art because
(he assumes) the relevant intentions were absent. A theory of error is provided
by the cluster account: primitive art is art, since intention is only a criterion,
not a necessary condition for something to be art. Adajian, as we saw, objects
that the situation is symmetric: the denitionalist can reply that the necessary
17

Adajian, On the Cluster Account of Art, pp. 383384.

BERYS GAUT

283

condition of intention is illicitly weakened to a criterion.18 However, this


would make a poor theory of error, as can be shown by considering less contentious examples. Artists practice sketches can be art, even though they
might not be intended as artworks, but be made merely as experiments or as
recordings of particular phenomena, to be worked up later into artworks. It is
highly implausible to deny that such sketches can be art, when the oil sketches
of artists such as Thomas Jones and John Constable are widely regarded as
being among their nest works. The point, then, is not merely to note that
there is a symmetry in possible explanations available to the cluster theorist
and the friend of denitions: the question is, which explanation is correct. And
here the cluster account wins.
Moreover, the cluster account is superior to at least simple denitions in
being able to explain the range of disagreements concerning what is art. The
friend of denitions can use his favoured essential characteristic to explain a
particular disagreement: for instance, if he denes art in terms of intentions, he
can try to explain disagreements that revolve around the presence or absence
of intention. But how is he to explain disagreements that revolve around the
presence or absence of other factors? For instance, in the case of conceptual art
the disagreement may centre on whether it is necessary for the putative art
object to have aesthetic properties (such as beauty) or to be the product of skill
by the artist. Since these properties play no role in his favoured denition of
art, it is mysterious why they should feature in the dispute at all. But for the
cluster theorist there is no mystery: each of these properties features as a criterion of arthood, so their relevance in the dispute is explained.
v. heuristic utility

I noted in Cluster that an account of the concept of art gains support if it ts


with true or promising theories about art, that is, if it has heuristic utility. The
cluster account holds that a wide variety of properties is relevant to somethings
being an artwork, and each of these properties is possessed by things that are
not artworks: hence the account ts smoothly with and fosters some of the best
work in aesthetics, work which has examined properties, such as expression
and representation, that have a wider application than simply to art.19
Adajian objects that denitions have just as strong a claim as the cluster
account to heuristic utility: expressive, aesthetic and formal properties are
18

19

In Art as a Cluster Concept, p. 37, I remarked that the cluster account has an explanatory advantage over simpler denitions in respect of explaining disagreements; more complex denitions, such as historical ones, I noted, might be able to generate an adequate
theory of error, albeit indirectly. (The same might apply to complex denitions ability to
provide a theory of error about the appeal of rival, simpler denitions.) However, Adajians
objection is that the cluster account enjoys no advantage over any denitions.
Art as a Cluster Concept, pp. 4041.

284

CLUSTER ACCOUNT OF ART DEFENDED

possessed by other domains besides art, so that denitions, such as expression


and formalist ones, are as well equipped to show interconnections with other
domains, and to connect to the philosophy of mind, language, and action.20
I do not deny, of course, that denitions can pick out properties that are
common to things which are artworks and things which are not. But the cluster account identies a wide variety of art-relevant properties, whereas denitions, such as aesthetic, formalist, or expression ones, standardly pick out only
one or two. In holding that there is a plurality of criteria in the cluster, most of
which can be shared by other domains, the cluster account increases the number
of connections between art and these other domains. So at least simpler denitions do not have just as good a claim as does the cluster account to showing
the degree of connections between art and other domains.
Secondly, the spirit of the search for denitions not infrequently displays
itself in seeking single properties that are unique to art. Here the position of
Clive Bell is emblematic: Bell identied the essence of art as signicant form,
which he held was unique to art; and this form is the cause and object of aesthetic emotion, which also is unique to art (not even natural objects have this
form or produce this emotion).21 The claim that there is a single, unique property that identies art is not required by the search for denitions (a denition
might hold, for instance, that there are two properties, whose conjunction is
unique to art, but which can individually be possessed by non-artworks). But
the search for denitions easily expresses itself in the search for unique properties. The spirit of the cluster account, in stressing the wide variety of art-relevant
properties, is alien to such an enterprise.
Finally, besides pointing out the interconnections that the cluster account
fosters with other domains, I noted in Cluster that in stressing the diversity
of art-relevant properties, the cluster account also meshes with pluralism about
aesthetic values and with a patchwork view of interpretation, which sees it as
ascribing a variety of distinct properties to works. This point remains untouched
by Adajians criticisms.
vi. denitions and the cluster account

Up to this point, I have been considering criticisms which hold that the cluster
account is no better than a denition of art, and which assume that the cluster
account is not a denition. The nal criticism is very different: it holds that the
cluster account really is a denition, and thus does not represent a radical
challenge to the correctness of denitions. Both Robert Stecker and Stephen
Davies point out that there are disjunctive denitions: indeed, Stecker holds
20
21

Adajian, On the Cluster Account of Art, p. 384.


Clive Bell, Art, ed. J. B. Bullen (Oxford: Oxford U.P., 1987), ch. 1.

BERYS GAUT

285

that all plausible denitions of art are disjunctive, and Davies gives examples of
other cluster accounts whose authors hold them to be denitions. Davies notes
that if one can say how many criteria have to be fullled for something to be
art (say, at least eight out of ten in my candidate account), then the cluster
account can be simplied into three disjuncts (either eight, or nine, or ten
criteria have to be fullled for something to be art). And in any case he sees
nothing in the account to show that the disjuncts are too numerous to qualify
as a complex, disjunctive, but otherwise orthodox, denition. Stecker believes
that as soon as any account species the content of the criteria, then we seem
to be back on the track of a denition and holds that the cluster account is tantamount to a denition if I would ultimately settle on a nite, denite set of
conditions in the cluster.22 Since I agree that this is the aim of the cluster
account, the cluster account would thereby count as a denition. Both authors
hold that the cluster account is a promising candidate for a correct account of
artindeed, for Stecker it is one of a group of denitions on which there is a
rough convergence in contemporary philosophy. Their objection is that it is
not what it purports to be: it is a denition.
The cluster account is not a denition in the sense in which I identied that
notion in Cluster, namely an account that gives individually necessary and
jointly sufcient conditions for something to fall under a concept; that is, it is
not a conjunctive denition.23 Let us suppose for the sake of argument that the
cluster account is a disjunctive denition: what would follow? We would need
to distinguish between simple disjunctive denitions, and highly disjunctive
and variegated denitions. The latter have a high number of disjuncts and the
conditions featuring in the disjuncts are signicantly different from each other.
In the candidate account, the criteria are indeed heterogeneous: involving skill,
possession of positive aesthetic qualities, being the product of a certain sort of
intention, being an exercise of creative imagination, and so on. Moreover, the
disjuncts are very numerous. In constructing the candidate account, since one
has to assess all selections from the ten criteria (assessing all possible disjuncts
with ten down to zero members), there are 1024 possible disjuncts to consider.
Only the disjuncts that sufce to make something art feature in the nal candidate account: but there are potentially very many of them. The resulting
account is plausibly a highly disjunctive and variegated denition. And the
number of disjuncts cannot be reduced by holding that a xed number of criteria have to be satised for something to count as art, for reasons that we saw
in discussing Duttons account. Simple disjunctive denitions, in contrast, have
very few disjuncts and/or the conditions in the disjuncts exhibit commonalities
22

23

Stecker, Is it Reasonable to Attempt to Dene Art?, quotations are from p. 47 and from
n. 11, p. 62; and Davies, The Cluster Theory of Art, quotation is from p. 299.
Art as a Cluster Concept, p. 27.

286

CLUSTER ACCOUNT OF ART DEFENDED

(they are not variegated). For instance, consider Steckers favoured denition:
very roughly, something is an artwork at time t just in case either it is in a central art form at time t and is intended to full a properly specied function of
that form, or it is an artefact that fulls a properly specied function of art with
excellence. On the surface, this is a simple disjunctive denition: there are
two disjuncts and there is a degree of commonality between themboth
involve appeal to function, hence the denitions neofunctionalist tag.24
We can now contrast highly disjunctive and variegated denitions, such as
the cluster account, with both simple disjunctive and conjunctive denitions.
We can then systematically recast the terminology of the position defended
here using this contrast: for instance, when previously it has been said that there
is no correct denition of art, this is now understood as holding that there is no
correct denition that is not highly disjunctive and variegated: conjunctive and
simple disjunctive denitions are awed. And the arguments for the cluster
account would still apply, but now recast as arguments for highly disjunctive
and variegated denitions, and against simple disjunctive and conjunctive ones.
So the SteckerDavies objection, even if it were successful, would require us
to recast our terminology, but would not undermine the substantive claims of
the cluster account. We would still be able to claim the superiority of the cluster account, by virtue of its highly disjunctive and variegated form, over (other)
denitions, and the arguments for that superiority would still stand.
However, there is reason to retain the anti-denitional terminology. An initial thought is that the denitionalists stand in danger of stipulating their terms
so that any clear and graspable account of a concept has to count as a denition of it. This danger is suggested by Steckers claim that when we specify the
content of the criteria in the cluster, we seem to be on the track of a denition. And he also writes, in the passage quoted earlier, that the cluster account
is tantamount to a denition, if it would settle on a nite, denite set of conditions in the cluster. But no correct account could employ an innite list of
disjuncts. We are trying to model a real human capacity (to apply the word
art), and that requires a nite list, if the list comprises variegated criteria. (The
innities that we can grasp are all structured, such as those of the cardinal
numbers, which are generated by recursion). So the danger is that the denitionalist position rests here on stipulation, in not allowing any clear, nitely
specied account of a concept to count as anything other than a denition.
The main reason to retain the anti-denitional terminology is that it locates
the dispute about the cluster account in its proper place in the history of
24

Stecker, Is it Reasonable to Attempt to Dene Art?, p. 47. I say on the surface, since
Stecker holds that there is a wide range of functions that art has had historically. Depending
on how wide and variegated they are, his denition might start to approximate a highly
disjunctive and variegated account.

BERYS GAUT

287

aesthetics since the 1950s: the cluster account is a prime candidate for what the
anti-denitionalist should hold, after she has subjected her position to reasoned
reection. The anti-denitionalist project is grounded, of course, in
Wittgensteins remarks about family resemblance. We need to disentangle the
resemblance-to-paradigm view in these passages from the cluster account
view, and adopt the latter, for reasons already mentioned. That account can
be given a disjunctive form, as Wittgenstein recognized; and it would be, in
our terminology, highly disjunctive and variegated: Wittgenstein writes of a
complicated network of similarities overlapping and criss-crossing that x the
meaning of game. The other aspect of his view is that these concepts do not
have xed boundaries, but this does not make them unusable, and that if one
wants to x a boundary, the only way is to draw it, rather than discover a preexisting boundary.25 So disjunctiveness and indeterminacy are the ingredients
that preclude the account from being a denition. Weitz picked up on both
these aspects (while failing to distinguish the resemblance-to-paradigm account
from the cluster account); but he laid particular stress on the latter, with his
famous claim that art is an open concept. It is because art is an open concept,
he thought, that one cannot dene it. A concept is open if
its conditions of application are emendable and corrigible; i.e., if a situation or
case can be imagined or secured which would call for some sort of decision on our
part to extend the use of the concept to cover this, or to close the concept and
invent a new one to deal with the new case and its new property.26

Since there are no predetermined boundaries, one has to make a decision as to


whether or not to include a new object as art.
Although the concept of art exhibits this kind of indeterminacy, I have laid
no stress on it, since plausibly all concepts exhibit this feature, so it does not
help in distinguishing those that can be dened from those that cannot. Weitz
himself holds that the only concepts that are closed are those of mathematics,
logic, and defunct practices such as extant Greek tragedy. Yet Wittgenstein, in
the passages to which Weitz alludes, suggests that mathematical concepts are
not closed. And the concepts of even defunct practices are not closed, since
we can imagine new works being discovered, and therefore received boundaries being called into question. Moreover, one can dene concepts that are
open. Take that hoariest example of a denition: a bachelor is an unmarried
man. The terms, married and bachelor require decisions about actual or
imaginable new cases. Is a man legally married if he goes through a marriage
ceremony, properly performed, but where, unbeknownst to all of the

25
26

Wittgenstein, Philosophical Investigations, part I, propositions 6669.


Weitz, The Role of Theory in Aesthetics, p. 15.

288

CLUSTER ACCOUNT OF ART DEFENDED

participants, the minister is not legally accredited? Is a woman who undergoes


a sex-change operation a male for purposes of marriage? These kinds of questions are the stuff of case law and legislative decisions. So a concept can be
open but denable by virtue of the openness of its dening terms.27
The upshot is that art is an open concept, but so are a plethora of other concepts,
many of which are denable. So the notion of openness will not differentiate
concepts that are denable from those that are not. Thus, if we are to capture, in
the light of reasoned reection, what the anti-denitionalists were after, we
should cast their anti-denitional claims not in terms of the idea of openness, but
in terms of the other aspect of their view: we should employ disjunctive accounts,
in particular highly disjunctive and variegated accounts. The cluster theory is
thus a prime candidate for being a reasoned reconstruction of the antidenitionalist position. As such it has good title to claim not to be a denition.
vii. conclusion

As I noted in Cluster, my aim is not to show beyond all doubt that art is a
cluster concept. What I have tried to show is that the standard story about the
denition of art has unjustly written-off cluster accounts. This is partly because of
confusions about their logical form, eliding resemblance-to-paradigm and cluster
accounts, and xation on the notion of an open concept. And I have also argued
that cluster accounts have signal advantages in directly matching our considered
intuitions about what is art, intuitions reected in the variety of factors that we
consider are relevant to whether something is an artwork. The cluster account
can answer the objections made to it, and though the question of whether it is to
be classied as a denition is not, I have argued, of fundamental importance, we
nevertheless have good grounds for denying that it is. Finally, besides defending
the form of the cluster account, I have also advanced a particular account, involving ten criteria. Interestingly, none of the cluster accounts critics discussed above
have criticized that particular candidate account. Yet it is in the development of
particular accounts and in a full specication of the disjuncts they involve that
the ultimate test of the cluster account will consist. What I have tried to establish here and in the previous paper is that this search for particular cluster
accounts is well worth undertaking, since they represent a most promising
avenue for those seeking to develop a correct account of the concept of art.28
Berys Gaut, Department of Moral Philosophy, University of St Andrews, Fife KY16
9AL, UK. Email: bng@st-andrews.ac.uk
27

28

Weitzs claim about the ubiquity of open concepts also raises a puzzle as to why he thought
that he needed an argument from creativity to show the openness of the concept of art, since
practices with open concepts are vastly more common than those that are creative.
I would like to thank Peter Lamarque for his helpful comments on this paper.

S-ar putea să vă placă și