GREGORY CURRIE
Actual Art, Possible Art, and Art’s Definition
It is thirty years since Jerrold Levinson introduced
his historical theory of art.1
Other, related theories
have been proposed.2
But Levinson’s theory is the
most elaborated and the most clearly committed
to the project of defining art.3
Much of the dispute
about this theory has centered on how precisely
it is to be formulated.4
I will be careful—to the
point, sometimes, of being tedious—to distinguish
ways of formulating the theory’s central claim. No
formulation I come up with provides a satisfactory
definition: so I argue.
I start by saying something absurdly brief about
the rules of engagement in this contest. I provide
some motivating background about biconditionals,
necessity, circularity, and their relation to the
project of definition. I characterize Levinson’s attempt
to avoid circularity by means of a technique
of collapse. I show that using collapse makes the
definition offered by the historical theory unacceptably
parochial. I suggest a way in which a
new historical definition might be crafted that is a
bit more cosmopolitan, though not, perhaps, cosmopolitan
enough. I also note that the historical
theorists need to take a stand on what seems to
me a difficult question concerning how we are
to interpret our intuitions about what would, in
counterfactual circumstances, be art.
i. method
Defining things is not to everyone’s taste. I shall
not take account of general doubts about the possibility
of or need for definitions. Even if we take
a generally upbeat approach to definitions, serious
doubts remain concerning Levinson’s proposal.
We do, however, need to ask whether the
project falls on one or the other side of a familiar
divide: that, roughly, between real and nominal
definitions. Is Levinson making a claim about the
nature of art or about the meaning of ‘art’? I believe
he is making a claim about the nature of
art.5
But there are signs that he also has his eye
on the semantic agenda, and there certainly are
connections between the two. To avoid a war on
two fronts, I am going to stick, as far as possible,
to the metaphysical question. How should we proceed
in thinking metaphysically about the nature
of art? An adequate metaphysics of art should be
responsive to how we intuitively think about art’s
nature and especially to how we think art might
or would have been different in different circumstances.
Without any such responsiveness we have
no way of keeping to topic. Our metaphysics may
violate intuition, but it needs a reason for doing
so. When I talk about our concept of art, I mean
something distinguished by intuitions of this kind;
people with different but fully reflective intuitions
about what would be art if the facts were arranged
this way or that way count as having different concepts
of art.6
Another way to put this is to say that
we have intuitions about what is art in various possible
but nonactual circumstances; call these intuitions
about possible art. At a certain point in the
argument, I will attend to Levinson’s claim that
he is concerned to make his definition conform to
our current concept of art. Consideration of our
intuitions about possible art show that claim to be
wrong; so I say. I will, however, issue a warning
about how difficult it can be to decide whether
people’s intuitive judgments in this area are fully
reflective.
The Journal of Aesthetics and Art Criticism 68:3 Summer 2010
c 2010 The American Society for Aesthetics
236 The Journal of Aesthetics and Art Criticism
ii. biconditionals, necessity, definition
Here is a purportedly necessary biconditional:
(1) Something is red iff it appears to be red to
normal subjects under normal conditions.
This is no reductive definition of ‘red,’ since the
term appears on both sides. Nonetheless, we may
hold that it is an informative proposition about
redness. It says that there is no more to something’s
being red than its looking red in the right
circumstances.7
(1) is said to be necessary. How is that to be
expressed? The most perspicuous way, given the
distinctions I shall need, is in terms of quantification
over worlds. For those worried by the metaphysical
strangeness or excess of possible worlds,
I say that use of this machinery to express modal
notions does not require us to believe in the reality
of possible worlds; I am doing it this way
because it provides a clear characterization of the
crucial differences between certain versions of the
historical theory. So we have:
(1ω) For all ω, something is red in ω iff it appears
in ω to be red to normal subjects under normal
conditions,
where ω ranges over worlds. This, of course, takes
us no closer to a noncircular definition.
Suppose we form a list, L, of all the things that
actually have, do, or will appear red to normal
subjects in normal conditions. We then say:
(1ωL) For all ω, something is red in ω iff it is
named on L.
It is important to see that the list is exactly that:
just a list of items. It does not tell us, for example,
that “these are the things that are red,” for then
we would not have got away from the circularity
of (1ω). The list specifies a certain set—the set of
things which are actually red—by specifying its extension.
It is that extensional specification which
appears on the right-hand side (RHS) of (1ωL).
We may say that (1ωL) employs a technique of
collapse: the intension on the RHS of (1ω) collapses,
in (1ωL), into an extension, getting rid of
the circularity.
We could never, of course, hope for a definition
along these lines because we could never
formulate the list. But even as formulated by an
omniscient being, (1ωL) would be a hopeless definition.
It says that exactly the same things are red
in every world, namely, the things on the list. (1ωL)
is maximally parochial: it purports to tell us about
the circumstances under which things would be
red and ends up claiming that, however circumstances
vary from those in the actual world, the
distribution of redness remains the same.
iii. a purportedly necessary biconditional
about art
As Levinson originally stated it informally, his
view is that “a work of art is a thing intended for
regard-as-a-work-of-art; regard in any of the ways
works of art existing prior to it have been correctly
regarded.”8
He has a good deal to say about when
something counts as having been intended for regard
in any of the ways prior works have been
correctly regarded, seeking to avoid objections to
his account based on unclarities concerning that
notion. I am going to assume that this is an entirely
unproblematic notion, so I will not review
any of Levinson’s elaborations on this—except to
note that he wants us to understand this relation
transparently: something counts as art so long as it
is intended for regard in way R, and R happens to
be a way in which some prior art was correctly regarded,
irrespective of whether the intender knew
this to be so. Also, it may happen that the artist
intends the work for regard in way R1, where R1
is an accepted way of regarding art. And later we
come to see that the work can usefully be regarded
in way R2, where R2 is not (yet) an accepted way
of regarding art. And later still, because of the
attention paid to the work, R2 may become an
accepted form of art-regarding. In that way, what
counts as a legitimate way of regarding something
as art changes over time.
Bearing this in mind, let us express Levinson’s
theory in this way:
(2) Something is art iff it is intended for regard as
some prior art was correctly regarded.9
What is the modal status of (2)? Art has, according
to Levinson, an “essential historicity,” and that
Currie Actual Art, Possible Art, and Art’s Definition 237
is certainly required if he is to tell us something
about art’s nature.10
That the ways appropriate to
regarding artworks at a given time include those
appropriate to regarding art at earlier times must
be a fact about art in any circumstances, actual
or not. As before, I will use quantification over
worlds to express this:
(2ω) For all ω, something is art in ω iff it is intended
for regard in ω as some prior art was
correctly regarded in ω.
According to (2ω), in whatever circumstances
we have art, we have something intended for regard
in the way that prior art was correctly regarded
in those circumstances.
Considered as a definition, (2ω) is circular in the
way that (1ω) is. This does not make (2ω) useless. It
might still be an important claim about the nature
of art, as (1ω) may be an important claim about the
nature of color. And if the project is a metaphysical
one rather than an attempt to give the meaning
of ‘art’ in other terms, why not be content with a
claim which displays (so it is claimed) art’s reflexive
essence?11
But Levinson goes to considerable
lengths to provide us with a noncircular definition,
apparently on the grounds that that is what someone
who wants to know the meaning of ‘art’ is
entitled to. At this point the metaphysical project
has taken on a semantic dimension. In line with
my earlier resolution, I will continue to treat this
as an exclusively metaphysical issue, which in part
it surely is. I now ask whether the purportedly
noncircular definition Levinson gives is adequate
as an account of the nature of art.
Levinson’s response to the circularity objection
is a shift parallel to the shift from (1ω) to (1ωL).
He suggests that (2ω) be reformulated in a way
which specifies what it takes to be art now, not
in terms of what is prior art, but simply by reference
to a list of regards, namely, those which, as a
matter of fact, have been, up to this point, legitimate
ways of regarding art. As Levinson puts it,
his account is not circular because “What it does
is define the concept: being art at a given time by
reference to the actual body of things that are art
prior to that time”—together, we should add, with
the list of ways these things actually were properly
regarded.12
Elsewhere he says, “the concrete [that
is the actual] history of art is logically implicated
in the way the concept art operates.”13
So now we
have:
(2ωL) For all ω, something is art in ω iff it is
intended for regard in ω in one of the ways
on L,
where L is simply the list of actual ways we have
of regarding art.14
iv. time and community
We must add two things to (2ωL) if we hope to
express the notion which Levinson is attempting
to capture. First, we do not count something as
art when it is put forward at time t for regard in
some way that is the regard intended for some
other things in the same community, but only at
some time later than t; that would make something’s
status as art now depend on future contingencies.
The regards relevant at t have to be those
acknowledged prior to t. Secondly, a world may
contain distinct, even isolated artistic communities;
a world is just too big an entity to suit all the
comparisons we need to make. But a community
is not just a bit of a world. It is a transworld entity,
for we can think about the same community in
distinct circumstances. So if K is a community, we
can speak about what is happening not merely in
K, but in K-in-world-W. Taking these two things
into account, we have:
(2ωLτκ) For all ω, τ, κ, something is art in ω at τ
for κ-in-ω iff it is made in κ-in-ω at τ, and
intended for regard by its maker in one
of the ways on L<τ,
where τ ranges over times, κ over communities,
and L<τ is that sublist of the list L that consists
of all the regards mentioned on L and operative
prior to τ.
v. a definition, but not of our concept of art
For all its complexity, the objection to (2ωLτκ) is
simple. Like (1ωL), it is maximally parochial: it
makes the art of every world at a given time, possible
or actual, depend very rigidly on what is prior
art for our community as it actually is. Various
objections to Levinson’s proposal have been, in
effect, attempts to show the parochialism of the
proposal. People have objected, for example, that
we ought not to make the capacity for art making
238 The Journal of Aesthetics and Art Criticism
of some alien society depend on our own art history,
of which they may know nothing. In responding
to such objections, Levinson has urged that
he is not attempting to define the concept of art
in general, but rather the concept of art which
we have at a given time. He says that “the concept
of art has certainly changed over time. . . . It
is thus worth emphasising . . . that my analysis is
aimed just at capturing what the concept of art is
at present.”15
Further, “insofar as anything outside
our art tradition is properly said to fall under our
concept of art, it is because we can appropriately
relate it to our tradition of art, and in particular
to the normative regards that have, as a contingent
matter of fact, emerged in that tradition.”16
This suggests that we should put aside art and focus
on art-for-us-now. By (double) instantiation
on (2ωLτκ) we get
(2ωLus,now) For all ω, something is art in ω for us
now iff it is intended for regard by its
maker in one of the ways L that is
available now,
where ‘us’ and ‘now’ are rigid designators of
communities and times: our community and the
present time.
I accept, for the sake of the argument, that
(2ωLus,now) picks out a legitimate category of
things: the things that are art for us now. Let us say
(again, for the sake of the argument) that someone
who understands and accepts (2ωLus,now) has
an adequate concept of art-for-us-now. Do they
thereby have a concept of art anything like the
one people actually operate with? No. The concept
of art-for-us-now is not the concept of art
that we now operate with. Let me explain.
Suppose we say:
(3) For all ω, something is now tall in ω iff it is
now of height greater than n in ω,
where n is some particular height, say 5 10 , the
height above which we currently count a male
as tall, and “now” rigidly designates the present
time. We might say that (3) defines the concept of
tall-for-us-now. Is that the concept of tallness that
we now have? No. We now understand that people
and things generally are tall when they are above
average height, whatever height that happens to be
for the relevant group at that time, and whatever
that time is. We understand that it is contingent
that a height above 5 10 counts as tall for human
males at this time. We understand that if human
nutritional history had been different in a certain
way—if some other world had been actual—then a
male above 5 9 would be tall. We understand that
a woman may be tall without being over 5 10 , and
that someone in another community may be tall
without being tall by our standards. Our present
understanding of tallness encompasses possibilities
that (3) entirely fails to recognize.
Turn now to art. Suppose we grant Levinson—
what is actually controversial—that our concept
of art is at least historically constrained: grant, in
other words, that it is an essential fact about art
that we judge current samples to be art on the basis
of their relations to our past artistic practice.
Just as we recognize the variability of the relevant
standards for tallness, we are able to recognize
the variability of the relevant standard for art. We
are able to recognize that the correct way to decide
whether some item of Martian ware is art is
to compare the intentions behind its making with
the ways of regarding available in the prior history
of Martian culture and to see that it would
be a mistake to say that these things are art only
if intended for regard in some way sanctioned by
us.17
We can see, additionally, that if our art history
had been different, and different standards
of regard had been sanctioned by it, then our current
standards for judging things put forward as
art now would be different. We can see that what
is to count as art in the future will depend on the
ways of regarding available to us at that time in the
future and not on those available now. We could
not allow for any of these things if our concept of
art were the one defined by (2ωLus,now).
Where does that leave us? Without a reductive
definition of art, certainly. But not, as I have
emphasized, without something worthwhile. Con-
sider:
(2ωτκ) For all ω, τ, κ, something is art in ω at τ for
κ iff it is made in κ-in-ω at τ, and intended
for regard by its maker in one of the ways
that art produced in κ-in-ω prior to τ was
properly regarded.
This is a circular biconditional, just a sprucedup
version of our earlier (2ω). But we have seen
that circular biconditionals can be significant indicators
of a thing’s nature. (2ωτκ) might stand as
an indicator of the nature of art, telling us that the
Currie Actual Art, Possible Art, and Art’s Definition 239
facts about the ways in which something now may
be put forward as art cannot outrun the facts about
the ways of regarding appropriate to prior art.18
That this is so is an interesting claim which would
survive the failure of an attempt at reductive definition.
There may be other reasons for regarding
it as false, but they are not my concern.
vi. giving due weight to the actual
This is not quite the end of the debate. A historical
theorist might agree that the definitional proposal
(2ωLus,now) ties the concept of art too rigidly to
the actual history of art. Still, she might say, there
is something in the idea of a connection between
art’s nature and the actual history of art. Our concept
of art, she says, is one which is to some degree
tied to the history of art we actually have, with
those ignorant of that history using the term ‘art’
deferentially, as most of us use ‘elm’ and other
natural kind terms.19
To that extent, art is somewhat
different from tallness; different enough,
anyway, to undermine the claimed parallelism exploited
above. (2ωLus,now) is too parochial, but
(2ωτκ) is too cosmopolitan. What we want is an
account of art’s nature which allows for some
counterfactual variation in art’s history, but not
for the unlimited amount enjoyed by things like
tallness.
With this idea in mind, we should move in two
stages. The first is to give a purely general characterization
of a historical practice of regarding.
Such a practice is one where items belong to the
current stage of the practice in virtue of being intended
for regard in ways that previous items in
the practice were. Such practices may be of many
kinds, since on no one’s account are all ways of regarding
to count as artistic ways of regarding. The
second stage is to specify those historical practices
of regarding, actual and possible, which are
to count as artistic, and hence as practices where
items belonging to the practice are works of art in
the historical sense. To this end we may say:
(4) A historical practice of regarding is artistic
if it is a historical practice of regarding that
is sufficiently similar to actual art-historical
practice.
Degree of similarity between actual arthistorical
practice and some other candidate arthistorical
practice, P, is to be measured by the
degree of overlap between the kinds of regards
on the list L and the kinds of regards in play in
P. If the overlap is complete, there is no difficulty
in counting P as an art-historical practice.
If the overlap is null, there are no grounds for
regarding P as an art-historical practice. What is
it for a historical practice to overlap sufficiently
with our actual art-historical practice? I do not
know, but I do not believe that this creates any
serious difficulty for the proposal, just as the fact
that I do not know how much hair someone must
lose to count as bald shows that there are no bald
people. We are dealing with something essentially
vague and perhaps contestable. People who agree
to take this approach to the idea of a historical
definition of art must expect to face cases they do
not know how to classify and perhaps to disagree
about cases without there being any way available
within the resources of the historical theory or perhaps
anywhere else to decide who is right. That
sort of vagueness and irresolvable disagreement
surely pervades our ordinary thinking about art
itself.
Still, I am not confident that this approach,
based on the overlap of two lists of regards, will
deliver the goods. First of all, it depends on resolving
a dispute between two different accounts
of our intuitive responses to cases of “alien art”
in a certain way—a dispute which strikes me as
hard to settle. The historical theorist claims that
we intuitively recognize a limited degree of dependence
on actuality in the nature of art itself. Why
say this, rather than saying that we are too much
dependent on actuality, and our experience of it,
for our knowledge of whether something claimed
to be art is or is not art? Why can it not be, in other
words, that we have a modally flexible concept of
art—(2ωτκ)—and much less flexible ways of knowing
whether something is art or not? The historical
theorist owes us a reason for thinking that it is art’s
nature which is dependent (to a degree) on actuality
rather than the quality of our judgments about
what is art.
Let us make this a little more concrete. Suppose,
in the spirit of experimental philosophy, we
come up with weird cases of communities of alien
beings who have traditions which involve patterns
of regarding things which have little or no overlap
with patterns of regard that we are used to applying
to the things we recognize as art. Assume
that, if we ask subjects in the experiment, “Are
240 The Journal of Aesthetics and Art Criticism
these things works of art?” they will say no. That
could be for either of two reasons. It could be because
they have a concept of art that tells them
that these things are contrary to the nature of art.
Or it could be because they simply think, wrongly,
that these things are not art. You might say that
the experiment is in fact decisive, because there is
no independent reason for thinking that subjects
would give a factually wrong answer. But there
is. Notoriously, people make category judgments
according to the prototypical features displayed
by stimuli, judging things that do not look like familiar
samples of water not to be water, and not
saying things like “Well, it doesn’t look like water,
but whether it is or not depends on its hidden
essence.” And this holds not only of natural kind
terms. If I show you something that looks nothing
like any carburetor you have ever seen and ask
you whether it is one, I fully expect you will say
no, even though in fact it is a carburetor, having
been very innovatively designed to fit the functional
role definitive of carburetors, and you understand
that carburetors are functional objects.
Similarly, it could be that people will judge these
weird objects to not be art simply because they do
not look sufficiently similar to familiar works of
art, and not because they have a restrictive understanding
of art’s nature. I do not despair of finding
ways to test between these two hypotheses about
the extent of counterfactual variation in art, but
I cannot see where, in advance of serious experimental
work, a historical theorist would get the
confidence to come down on one side rather than
the other.
Here is another reason for doubting the acceptability
of (4). Imagine a historical practice of regarding,
P, which starts off pretty much as our arthistorical
practice of regarding did (however that
was), but thereafter diverges in such a way that
the list of regards associated with that practice has
minimal overlap with L. P might have a history
which makes the temporal shift in ways of regarding
within that practice perfectly intelligible as an
artistic practice: we might be able to provide a
narrative which links the stages of P in such a way
that each change in regard is an intelligible result
of a response by some practitioner to the kinds
of regards previously available. Each change in
the practice takes it further away from the actual
art-historical practice we are supposed to measure
it against, but each change, considered in its own
terms, seems like a perfectly legitimate, perhaps
even highly creative, artistic move.20
Robert Nozick
once argued that a current distribution is just if
it is obtained from a just distribution by changes,
each of which is just.21
I take no view on whether
Nozick’s principle is right. But I am suggesting that
a historical theorist ought to accept an analogous
principle: that a current practice of regarding is
artistic if it is obtained from an artistic practice of
regarding by changes, each of which is intelligible
as an artistic response to that prior practice.
In the face of such an objection, the historical
theorist could move up a level: replacing, or
supplementing, emphasis on overlap in ways of
regarding with emphasis on overlap in ways of
shifting between ways of regarding. A historical
practice of regarding would then be artistic to the
extent that its ways of shifting between ways of regarding
overlap with the ways exemplified in our
own art history. Or perhaps a measure of artisticness
would be given by some weighted average
of these two kinds of overlap. I leave it open as
to whether historical theorists will be able to develop
this suggestion. If they are not, or if they
are unable to resolve in their own favor the choice
between competing explanations of our intuitions
about what is possible art, I invite them to fall back
to (2ωτκ) and to claim that it offers an illuminating,
if nonreductive, thesis about the nature of art.22
GREGORY CURRIE
Department of Philosophy
University of Nottingham
Nottingham NG7 2RD
United Kingdom
internet: gregory.currie@nottingham.ac.uk
1. “Defining Art Historically,” The British Journal of
Aesthetics 19 (1979): 232–250; “Refining Art Historically,”
The Journal of Aesthetics and Art Criticism 47 (1989): 21–33,
both reprinted in Jerrold Levinson, Music, Art and Metaphysics,
2nd ed. (Oxford University Press, 2006); “Extending
Art Historically,” The Journal of Aesthetics and Art Criticism
51 (1993): 411–423, reprinted in The Pleasures of Aesthetics
(Cornell University Press, 1996); “The Irreducible Historicality
of the Concept of Art,” The British Journal of Aesthetics
42 (2002): 367–379, reprinted in Contemplating Art
(Oxford University Press, 2006). References throughout are
to the reprints.
In formulating my objection to Levinson’s proposal, I
am drawing heavily on the objections of a number of authors:
Stephen Davies, Berys Gaut, Graham Oppy, Robert
Stecker, and others. My contribution is at most one of
overall clarification. For a comprehensive bibliography of
Currie Actual Art, Possible Art, and Art’s Definition 241
contributions to the debate, together with the locations of
Levinson’s various replies, see his “The Irreducible Historicality
of the Concept of Art.”
2. For example, James D. Carney in “Defining Art,” The
British Journal of Aesthetics 15 (1975): 191–206; “What Is
a Work of Art?” Journal of Aesthetic Education 16 (1982):
85–92; “Defining Art Externally,” The British Journal of
Aesthetics 34 (1994): 114–123; and No¨el Carroll in “Historical
Narratives and the Philosophy of Art,” The Journal
of Aesthetics and Art Criticism 51 (1993): 313–326,
reprinted in No¨el Carroll, Beyond Aesthetics (Oxford University
Press, 2001); see also other essays in Part II of that
volume.
3. In a recent essay, Levinson refers to his account as
“the sort of complete definition of art I am inclined to favor”
(“The Irreducible Historicality of the Concept of Art,”
p. 13).
4. Levinson recently rejected an attempt of my own to
formulate it. See my “A Note on Art and Historical Concepts,”
The British Journal of Aesthetics 40 (2000): 186–190,
and Levinson’s reply in “The Irreducible Historicality of the
Concept of Art,” pp. 22–25.
5. Introducing his theory, Levinson said, “What is the
artness of an artwork? Wherein does it reside?” (“Defining
Art Historically,” p. 3). But see below, note 11, and accompanying
and following text.
6. I do not claim, and do not need to claim, that sameness
of intuition makes for sameness of concept. I also do
not claim that this is anything more than one way, useful
in certain circumstances, to distinguish those mental competencies
which have some right to be thought conceptual.
I do not claim that other theories of concepts and concept
possession, of which there are radically different varieties,
are wrong.
7. At least, it can be interpreted this way, rather than as
a statement to the effect that the perceptual sensitivities of
normal subjects in normal conditions track the facts about
color. For an introduction to the complex issues raised by this
distinction, see, for example, Crispin Wright, “Realism: The
Contemporary Debate—W(h)ither Now,” and Mark Johnston,
“Objectivity Refigured: Pragmatism without Verificationism,”
both in Reality, Representation and Projection, ed.
John Haldane and Crispin Wright (Oxford University Press,
1993), pp. 63–84 and 85–130, respectively.
8. Levinson, “Defining Art Historically,” p. 6. A statement
roughly equivalent to this is given in Levinson’s “The
Irreducible Historicality of the Concept of Art,” p. 6.
9. Levinson warns us that, because of the problem of
revolutionary art, we may have to vary the right-hand side of
(2) along these lines: “X is intended for regard as some prior
art was correctly regarded or in some other way in contrast
to and against the background of those ways” (“Defining
Art Historically,” p. 17). For the sake of simplicity I will
ignore this.
10. Levinson, “Defining Art Historically,” p. 4.
11. At pp. 14–15 of “Defining Art Historically,” Levinson
manifests, I believe, some uncertainty as to whether and
if so in what respect the circularity objection really is an
objection to his project. He says that “to eliminate this reflexivity
would be to eviscerate the term ‘art,’” but goes on to
claim that the definition he offers does eliminate reflexivity.
12. Levinson, “Defining Art Historically,” p. 14.
13. Levinson, “The Irreducible Historicality of the Concept
of Art,” p. 24, emphasis added.
14. There are other ways to interpret what Levinson says
about this. One is to take the use of ‘actual’ in “actual body of
things that are art” as logically redundant but there merely
for emphasis. That would be to say, in effect, that what determines
whether something is art now in a world W depends
on the prior history of art in W—W being the actual world
from the point of view of W itself. This would amount to
going back to the circular (3ω). Alternatively, we could understand
the proposal in this way: instead of thinking of each
possible situation as given by a possible world, we think of
possibilities as given by a choice of one world as counterfactual
and another as actual. In that way we can think about
what would be art in a given (counterfactual) world if some
other world were the actual world, and we would get different
answers to the question, “What is art in world W?”
depending on the choice we make about which other world is
actual. This amounts to treating Levinson’s proposal within
what is called two-dimensional semantics. The prospects for
such an approach were discussed in the papers referred to
in note 4 above. While I continue to believe that the twodimensional
approach to this issue has merit, it does not, I
think, offer any route to the kind of noncircular definition
of art that Levinson wants. So for present purposes we need
not consider it.
15. Levinson, “Defining Art Historically,” p. 23.
16. Levinson, “The Irreducible Historicity of the Concept
of Art,” p. 21. The first emphasis is mine.
17. This seems to be acknowledged by Levinson, who
approves a very clear statement of the position concerning
Martian art given by Stecker. See Levinson, “The Irreducible
Historicality of the Concept of Art,” p. 373n20,
quoting from Robert Stecker, Artworks: Definition, Meaning,
Value (Pennsylvania State University Press, 1997),
p. 108.
18. Here I adapt a formulation of Mark Johnston’s concerning
color (“Objectivity Refigured: Pragmatism without
Verificationism,” p. 106).
19. I am not suggesting that ‘art’ is a natural kind term:
merely that it is a term used deferentially, as natural kind
terms are.
20. I am indebted here to No¨el Carroll’s idea that later
activity counts as artistic if there is a narrative which intelligibly
links it to earlier artistic practices and in the light of
which the later activity can be seen as a response to problems
intrinsic to the earlier artistic practice. Carroll emphasizes
that he is not offering a definition of art. See his “Historical
Narratives and the Philosophy of Art.”
21. See Robert Nozick, Anarchy, State, and Utopia (New
York: Basic Books, 1974), p. 151.
22. I am grateful to Jerry Levinson for comments on an
earlier and significantly different version of this article, and
for a suggestion about the title.