Intentionalism in Aesthetics

Paisley Livingston
Intentionalism in aesthetics is, quite generally, the thesis that

the artist's or artists' intentions have a decisive role in the creation

of a work of art, and that knowledge of such intentions is a necessary

component of at least some adequate interpretive and evaluative claims.1
In this paper I develop and defend this thesis. I begin with a discussion
of some anti-intentionalist arguments. Surveying a range of intentionalist
responses to them, I briefly introduce and criticize a fictionalist version

of intentionalism before moving on to an approach I call moderate

intentionalism. I consider a salient alternative known as hypothetical
intentionalism and try to show why moderate intentionalism should be
preferred to it.

Saying what, precisely, intentions are is no small problem, and

disputes in aesthetics often hinge on rival assumptions about the nature
and function of intentions in general. I shall assume, in what follows,

that intentions are mental states having semantic contents, various

psychological functions, and practical consequences?but not always the
targetted results.21 shall not take up any of the more global challenges
to intentionalist psychology, such as eliminative materialism or macro
sociological and historicist critiques.3 I assume, then, that agents some
times intend to perform an action, such as writing a poem, and that they
occasionally succeed in realizing such aims, thereby intentionally doing
such things as writing poems.

I. Extreme Intentionalism and Anti-Intentionalism

In an extreme version, intentionalism holds that a work's meanings
and its maker's intentions are logically equivalent. Such a thesis still has
its defenders, yet it is hard to see how it can be reconciled with the fact
that intentions are not always successfully realized. A theory of interpre
tation based on Humpty Dumpty's semantics does not seem promising.4
An extreme version of anti-intentionalism also has its advocates, who
confront the intentionalist with the following dilemma: either the artist's
intentions are successfully realized in the text or structure produced by
the artist, in which case the interpreter need not refer to them; or, the
New Literary History, 1998, 29: 831-846

This content downloaded from 59.189.72.200 on Tue, 31 Jan 2017 10:12:28 UTC
All use subject to http://about.jstor.org/terms

832

NEW LITERARY HISTORY

artist's intentions are not successfully realized, in which case reference

to them is insufficient to justify a related claim about the work's

meanings. Any viable form of intentionalism must find a way out of this

dilemma.

A premise of many anti-intentionalist arguments?including the di

lemma just mentioned?is that if a work has determinate meanings and
value, they must be immanent in the artistic text or structure. This sort
of empiricism in aesthetics is vulnerable to some powerful criticisms.
Not all of the artistically or aesthetically relevant features of a work of art

are intrinsic properties of the text; some are relational and can only be
known when the text or structure is cognized correctly in the context of

its creation. In making this point, a number of philosophers, such as

Arthur Danto, David Davies, Jerrold Levinson, and Gregory Currie, have

evoked versions of Jorge Luis Borges 's fictional example of Pierre

Menard: tokens of the same text-type, created in different contexts,
manifest different, artistically relevant relational features; to know which
features are those of one work as opposed to another work, one must
interpret the text in its context of creation.5
Once attention has been drawn to the constitutive status of a work's

relational properties, cogent responses to the anti-intentionalist di

lemma can be formulated. The intentionalist can argue that some
successfully realized intentions are not simply redundant with regard to
the text's intrinsic features. An example is the intention that a certain
meaning be unstated in the text yet implicitly expressed by the work.
Even when the intentions are successfully realized, such relations are not
immanent in the final artistic structure or text and cannot be simply
read off from the latter. Intentionalists also contend that whenever our

goal is to evaluate a work as a certain kind of achievement, the artist's

intentions, including unsuccessfully executed ones, are always relevant,
because part of what we want to do is take note of the manner and extent
of the artist's realization of the relevant aims. Although it is not the case

that success at realizing one's intentions entails success at creating a

valuable work, success or failure in realizing intentions does have
implications for the kind of value a work possesses, if only because there

is a significant and relevant difference between lucky and skillful

creative activities. No?l Carroll contends, for example, that Ed Wood's

Plan 9 From Outer Space would have been a better film had the director
been trying to make a parody. What one sees and hears at a screening of

this movie is logically compatible with such an intention, but our

knowledge that the director in fact had no such intention is decisive,

and we cannot justifiably praise the film as a clever parody.6 So the

dilemma can be avoided: knowledge of the relation between an artist's

intentions and the resulting structure is necessary to at least some

interpretive and evaluative claims.

This content downloaded from 59.189.72.200 on Tue, 31 Jan 2017 10:12:28 UTC
All use subject to http://about.jstor.org/terms

INTENTIONALISM in aesthetics

833

One way in which anti-intentionalists challenge even modest versions

of intentionalism is to evoke epistemological worries about the difficulty
or impossibility of obtaining reliable knowledge of intentions. Usually
this amounts to an unjustified demand for a kind of infallible justifica

tion or proof that is unattainable in any empirical domain, and no

reason is given why such high standards should be imposed on claims

about intentions and other mental states. The inconsistency is flagrant
when the theorist who voices such skepticism about intentions makes all
sorts of bold claims about such complex topics as the nature of textuality
and the ways in which readers construct authors. If we can know how
readers construct things, why can we not know how artists do so?

A more reasonable worry voiced by anti-intentionalists is that

intentionalist strictures make the focus of appreciation shift radically

away from the text or structure and toward the life and mind of the
artist, the ultimate result being a ratification of the most narrow variety
of biographical criticism. Although the latter has its rewards, it does not
yield the only or the most valuable interpretations of an artist's works.

For example, George Painter's biographical study of Marcel Proust

certainly sheds some light on Proust's achievement, but interpretations

by Vincent Descombes, Ren? Girard, and many others are also valuable,

partly because they elucidate Proust's texts in ways that the writer

himself would never have done so.7 So if we are to defend some form of

intentionalism, it should not be one that prescribes only biographical

approaches or one that prohibits novel and creative interpretations,

including those that explore the significance that a work takes on
outside the context of its creation. Part of the solution, then, is to
observe that interpretations can manifest many different sorts of value.

Elucidating the meanings and artistic value a work had in its original
context is one sort of valuable goal, but finding clever, new, and even
anachronistic ways of using a text can also be worthwhile. It is not clear
that these two sorts of interpretive projects are always in competition

with each other. So a premise of any tenable intentionalist theory of

interpretation is that a valued but not exclusive goal of interpretation is
the epistemic one of knowing something about the work of art qua work
of art, and this in its original context of production. The intentionalist
theses I discuss in the rest of this paper all share this premise.

IL Fictionalist Intentionalism
One approach that appeals to many critics is to maintain some sort of
intentionalist framework while adopting an instrumentalist stance with

regard to authorship. Looking at the textual evidence relevant to an

artistic corpus, the interpreter seeks, then, not to build the most realistic

This content downloaded from 59.189.72.200 on Tue, 31 Jan 2017 10:12:28 UTC
All use subject to http://about.jstor.org/terms

834

NEW LITERARY HISTORY

possible portrait of the life and works in combination, but to yield an

interesting and rewarding interpretation of the works, viewed as the
product of a postulated or fictional author. Such an approach has been
defended by Alexander Nehamas, for whom the target of interpretation
is the attitudes of a non-existent, make-believe authorial figure.8

The sort of fictionalist intentionalism is awkward, I think, because

there is an unresolved tension between two tendencies. On the one
hand, the interpreter is supposed to be building a picture of an agent
whose actions and attitudes explain the genesis of the work, and such a

process is presumably governed by constraints having to do with

psychological plausibility. If we really are interested in how works are

made, we should be interested in the desires, beliefs, intentions, and

other relevant attitudes of the actual makers, and we should also deem it

relevant to know whether an attitude has been intentionally expressed

or has only accidentally been made manifest in the work. Yet in spite of

its apparent emphasis on artistic agency, fictionalist intentionalism

describes an interpretive process that is not really aimed at forming a
hypothesis, not even a selective one, about the actual writer; instead,
interpreters only imagine or make-believe that the fictional entity they
describe was responsible for creating the text. As a result, significant
differences between possible creative histories are effaced.
Consider, for example, cases where an interpreter holds, solely on the

basis of evidence that is in principle accessible to members of the

appropriate audience, that there are artistically significant implicit pat
terns and meanings in a novel. The history of the work's production
could have been a matter of three different kinds of processes: (1) the
implicit meanings could have been intended by the artist, who wanted
them to be implicitly conveyed by the work; (2) the author may have had
no such intentions, but ended up writing a text compatible with such a

reading; or (3) the author may have had such relations in mind,

intending to make them explicit in the text, but failed to realize this
intention. The fictionalist intentionalist who is attuned to the implicit
pattern cannot speak of the differences between these three cases, and
can only describe the implicit meanings as expressing the intentions of
a fictionalized author. The actual author's intentions, when successfully

acted on, do not have any constitutive or evidentiary role in making

these implicit relations part of the work's artistic content.

Textual appearances can be deceptive: a text that emerges from a

chaotic and uncoordinated process of multiple authorship involving the

efforts of various individuals could look like it resulted from this kind of

messy history; but on the other hand, it could look like it had been
intentionally produced by a single author or group of authors acting on

a reasonably well-conceived and executed scheme.9 Similarly, a text

This content downloaded from 59.189.72.200 on Tue, 31 Jan 2017 10:12:28 UTC
All use subject to http://about.jstor.org/terms

INTENTIONALISM in aesthetics

835

intentionally crafted by a single (or collective) author acting on a well

conceived and well-executed scheme could look like something emerg
ing from an uncontrolled process of multiple, uncoordinated contribu
tions; but it could also look like some author's controlled, intentional
doing. The interpreter who is oblivious to the intentional or uninten
tional nature of the actual creative process, as opposed to the text's
appearances, is not in a position to distinguish between these four
different kinds of cases. Working with a default assumption in favor of
intentional creation and expression, the fictionalist runs the risk of
mistaking accidentally coherent textual meanings for intentionally ex
pressed attitudes, attributing the latter to a nonexistent creator.

III. Moderate Intentionalism

Moderate intentionalism is the thesis that the actual maker(s)' atti
tudes and doings are responsible for some of a work's content, and as
such are a legitimate target of interpretive claims; more specifically,
knowledge of some, but not all intentions is necessary to some, but not
all valuable interpretive insights because such intentions are sometimes
constitutive of the work's features or content. Moderate intentionalism
recognizes that the artist's intentions do not always constitute the work's
meaning. The contention, rather, is that when intentions are compatible
with the text, they can be constitutive of a work's implicit meanings. Just

as hinting and insinuation are part of the pragmatics of everyday

conversational exchange, so do artists sometimes enhance the value of
their works by expressing attitudes in an implicit and indirect manner.
In many artistic contexts, subtlety is a valuable feature, and bluntness a

failing.
Moderate intentionalism's claims about the implicit meanings of a
work can be articulated within a broadly Gricean framework where the
notion of conversational implicature has been adapted so as to develop

a conception of what could be dubbed "artistic" implicature.10 A key

claim, then, is that appropriate inferences made within the artist/

interpreter relation are guided by assumptions analogous?but certainly

not identical?to the maxims proposed by Paul Grice with reference to

everyday conversation. Artistic implicatures, then, are inferences to

implicit content based on the explicit content of a text or artifact, as well

as on assumptions shared by artists and their audiences, including

contextual beliefs and beliefs about the nature of the artist/interpreter
interaction. For example, authors and interpreters are guided by the

hypothesis of a "thin" authorial rationality: if an author intends to

express p implicitly, the author will try to adopt expressive means that

This content downloaded from 59.189.72.200 on Tue, 31 Jan 2017 10:12:28 UTC
All use subject to http://about.jstor.org/terms

836

NEW LITERARY HISTORY

are likely to make p manifest to interpreters who are reasonably

competent at assessing textual and contextual evidence. To that end, the

author intends to write a text that does not contain p as part of its
explicit content, a text, however, which will make it possible (if not
highly likely) for the members of the audience to infer the implicit
content by relying on both the text and contextual assumptions. What is

more, when authors try to communicate something implicitly, they

sometimes intend for their success in realizing this aim to depend on the
audience's recognition of that intention. A schematic illustration of how
moderate intentionalist principles may be exemplified in the interpreta
tion of a work of fiction is provided below in section V.

IV. Hypothetical Intentionalism

Moderate intentionalism is, I think, the right way to go, but it is
important to see how one may defend it against certain challenges. One
objection that appears frequently in the literature takes the following
form. Take some literary text or artistic structure that is well known and
that is generally recognized as having valuable and complex meanings,

and imagine that we were to discover that the artist in question

produced the work while acting on only some very limited semantic
intentions. Does not moderate intentionalism then have the crippling
consequence of requiring us to limit our understanding of the work's
original, artistic meanings to the ones intended by the artist? And why
should we want to do this when the interest and value of the work would

appear to suffer as a result? In Jerrold Levinson's version of this

challenge, we are asked to imagine that we discover that Franz Kafka's

intentions with regard to "Ein Landarzt" were simply a matter of

critiquing rural medical practices. Should we not reject any hermeneu

tic principle that would have the deflationary consequence of forcing us
to ignore the rich symbolic dimensions of Kafka's story?11
The idea behind this sort of challenge and the examples that are used
to illustrate it is that some artist's intentions can detract from the value
or interest of a work, and that a theory of interpretation should provide
a principled way of ruling them out. An approach along these lines is
ably defended by Levinson, who follows William Tolhurst in speaking of
hypothetical intentionalism.12 Crucial to this approach is a distinction
between two main kinds of intentions, labeled "semantic" and "categorial"
by Levinson. Speaking quickly, Levinson proposes that categorial inten
tions can determine a work's features and therefore have a constitutive
status, while semantic ones cannot and are at best suggestive of a work's

meanings. Whenever heeding someone's semantic intentions would

This content downloaded from 59.189.72.200 on Tue, 31 Jan 2017 10:12:28 UTC
All use subject to http://about.jstor.org/terms

INTENTIONALISM in aesthetics

837

make the interpretation less interesting and the work less valuable, we

should overrule them in favor of a superior interpretation that is

compatible with the textual and contextual data. In the case of the Kafka
example, Levinson's strictures would have us rule out Kafka's inferior
semantic intentions while retaining our crucial knowledge of his larger

categorial aims.

Such an approach obviously hinges on the distinction between

semantic and categorial intentions, which is drawn, first of all, on the

basis of the contents of intentions, and involves, more specifically, the
aspects of the work of art that the artist has in mind. As Levinson puts it,

categorial intentions "govern not what a work is to mean but how it is to

be fundamentally conceived or approached" (1188). In one of Levinson's
examples, the intention to make a sculpture and have it be taken as such
is categorial, while the intention to express rage with this work of art is
semantic.13
Why should categorial intentions have a different status in a theory of

interpretation? In what follows, I survey various potential reasons,

contending that on closer inspection, they do not in fact justify the use
made of the distinction in hypothetical intentionalism.
One potential reason for a difference in status has to do with reliability.
Perhaps the two kinds of intentions have significantly different functions

in the creative process, in which case interpreters who follow the

principles of hypothetical intentionalism are attuned to an important

difference. Levinson writes that categorial intentions are decisive or

determinative of a work's features in a way that semantic intentions are
not. He points out that semantic intentions often fail?as a result, say, of
clumsiness or mistaken beliefs. He then adds: "But if the writer intends

his text as a poem?as opposed to a short story, a dramatic monologue, a

piece of calligraphic visual art, or a mere diary entry?then that

intention is of a different sort and of a different order, and virtually

cannot fail?so long as the text in question at least allows of being taken,

among other things, as a poem" (I 188). In the same context, Levinson

goes on to say that semantic intentions do not "determine" meaning,
while categorial intentions "do in general determine how a text is to be

conceptualized and approached on a fundamental level and thus

indirectly affect what it will resultingly say or express" (I 189). And that,
presumably, is a reason, perhaps even a sufficient reason, why semantic

intentions should have only a suggestive role in the construction of

interpretive hypotheses, while categorial ones have an "evidential role."

Levinson says semantic intentions can fail; categorial ones virtually
cannot fail. Does this phrase mean that they do, sometimes, fail? So it

would seem. Levinson allows that both categorial and semantic inten
tions are fallible, so the reliability of the former is no reason for granting

This content downloaded from 59.189.72.200 on Tue, 31 Jan 2017 10:12:28 UTC
All use subject to http://about.jstor.org/terms

838

NEW LITERARY HISTORY

the two a logically distinct status in our theory of interpretation. Perhaps

Levinson's point in this regard is that it is in general easier to realize

categorial intentions, and that semantic intentions are more likely to
misfire. Yet even this more modest thesis is not so obvious. Some
categorial intentions may, in some contexts, be very hard to realize; and
some semantic intentions are easy to pull off. Degree of difficulty and
likelihood of success do not in any case correspond in any simple way to

whether knowledge of someone's aims has constitutive or merely

suggestive value with regard to their actual achievement.
A version of the intentional fallacy pertains to categorial intentions

just as much as it does to semantic ones. We cannot infer from

someone's having a categorial intention that it has been successfully

realized in the work, even if the agent is known to have acted on that

intention. Nor can we automatically infer back from features of a

realized text or artistic structure to the relevant categorial intentions. A
writer shows us a sonnet he has authored. Can we conclude, therefore,
that the author categorially intended to write a sonnet and intentionally
did so? The argument is invalid, even if we are willing to set aside cases

of wayward causality. The author could have been trying to realize a

specific categorial intention incompatible with the poem's actually
being a sonnet.
Levinson allows that categorial intentions are decisive or determinant

only if the text "allows of being taken" that way. The same sort of
constraint can be placed on our use of facts about semantic intentions:
a semantic intention is to be deemed decisive of a work's content only if
the text "allows of being taken" that way. Semantic intentions do not,
indeed, succeed "by fiat," but neither do categorial ones. In both cases,
recognition of the artist's constitutive role is constrained by facts about

what the writer has managed to do in producing a text. Moderate

intentionalism can make use of the same insight, holding that intentions

of any stripe are decisive only if they are textually or structurally

compatible, that is, if they are consistent with the features of the work's
text or artistic structure. Successful realization of intention is, in both

cases, a matter of intentionally producing something compatible with

the content of the intention.14

Reliability, or degree thereof, turns out not to be the key to any

important difference in status between the two kinds of intention, and
therefore not a decisive reason for preferring hypothetical intentionalism
over moderate intentionalism. Are there other reasons? One candidate

to consider has to do with accessibility or epistemic access: perhaps

categorial intentions are more readily known, while semantic ones are

elusive. Yet once we are in the business of making claims about the
respective contents of intentions, we are in no position to say that

This content downloaded from 59.189.72.200 on Tue, 31 Jan 2017 10:12:28 UTC
All use subject to http://about.jstor.org/terms

INTENTIONALISM IN AESTHETICS

839

semantic intentions are inscrutable creatures of the mentalistic night,

while categorial ones are solid and scrutable features of objective

behavior. We manage to know both sorts of intention?when, that is, we

do manage to know them at all?in the same way. Some categorial

intentions are, in any case, very hard to fathom. What precisely was the

categorial intention of Apuleius when he wrote the last part of The

Golden Ass? To write a parody, or something else entirely? What, exactly,

were Virginia WoolFs categorial intentions when she wrote Orlando?

Which of the passages in Franz Kafka's notebooks were meant to be just
diary entries, and which parts were intended to be works of literary
fiction? What were his categorial intentions when he interrupted his
writing to draw sketches in these notebooks? Did these three writers

successfully realize their categorial intentions, whatever they were?

Puzzling questions, as Baudelaire sometimes said.
Another unjustified asymmetry in the treatment of the two kinds of
intentions concerns their implications for a work's value. If a semantic
intention would lead to a lower estimate of the work's value, it should be

disregarded. Yet the hypothetical intentionalist does not think that

unfortunate or unsuccessful categorial intentions should be similarly
replaced. For example, the hypothetical intentionalist does not agree to

replace Ed Wood's serious categorial intent with a more appropriate

parodie one, thereby improving on his film. Why not? Perhaps because
it is important to recognize that it is the artist who creates the work and
its artistic value, as opposed to the critic or theorist. It seems plausible to
think that the latter premise should apply to semantic intentions as well.
Another problem with hypothetical intentionalism is that the distinc
tion is not sharp enough to be used the way this theory of interpretation
prescribes. Can we in practice sort out the contents of various intentions
involved in the making of works of art, deciding that some are categorial
and others not, and giving constitutive status only to the former? It looks
like there are some clear-cut cases where the rule can be applied fairly
easily, but a lot of cases where it cannot.
The intention to make a work of art as opposed to something else may
be an example of a clear-cut case, assuming we have a successful analysis
of the art/nonart distinction and of the place of intentions therein.
And, with regard to at least some of the more specific art forms, the
same could be true. It seems plausible, for example, to say that an artist
can have the intention of making a sculpture, even a representational

one, without having any idea what the sculpture will represent. He
intends to figure that out once he has his hands in the clay. We may even
want to say that he categorially intends to develop his semantic inten
tions later, in which case the categorial intention is a "second-order"
intention quite unlike a first-order semantic one.15

This content downloaded from 59.189.72.200 on Tue, 31 Jan 2017 10:12:28 UTC
All use subject to http://about.jstor.org/terms

840

NEW LITERARY HISTORY

Are there semantic intentions devoid of categorial contents? Perhaps.

Stirred by a powerful emotion of rage, the artist intends to express it, but
has no other intentions with regard to how this will be done. Like Pier

Paolo Pasolini, he is very versatile. Perhaps he will make a movie;

perhaps he will write a political poem, or maybe he will make a sculpture

depicting some angry workers. It could be, though, that the intention to
make a work expressing rage must involve at least a negative categorial

component: intending to communicate a sense of outrage, the artist

cannot intend to make a purely decorative or minimalist work. A sonata

for dog whistle probably will not do.
Even if we allow that some intentions can be sorted nearly, the friend
of hypothetical intentionalism must help us deal with the messy cases,
and I suspect there are a lot of them. Consider the intention to make a
work of fiction (as opposed, say, to some sort of nonfictional work). On

some prominent accounts, the intention to have members of an

audience recognize that a text or structure has a certain propositional

content is a necessary component of the larger fiction-making inten
tion.16 If such analyses are correct, the relevant intention is a mixed

affair, having categorial and semantic components. Or consider a

writer's intention to write a novel belonging to a trilogy, where the

intention, more specifically, is to create various meaningful, implicit

relations between the characters in the three novels. The writer intends

for the readers to think about the successive protagonists "as if they
were continuations of a single type of person.17 It seems hard to separate
the categorial and semantic aspects of the content of such an intention,
or cluster of interrelated intentions. Can we truly isolate the artist's goal
of making works that are part of a trilogy from the meanings that in his
mind constitute the links between the stories related in the three works?

Given such cases, how are we to apply an interpretive principle that

instructs us to exclude "any fact about the author's actual mental state or

attitude during composition, in particular what I have called his

semantic intentions for a text" (I 206)? Perhaps any intention having a
categorial component should be recognized, even if it is "contaminated"

by semantic elements. Or should we, on the contrary, decide that a

categorial intention involving a semantic component is merely "sugges

tive" and not constitutive? Both solutions seem arbitrary. The latter
sacrifices important categorial intentions in order to screen out the
semantic intentions; the former violates the clause about excluding
semantic intentions. Moderate intentionalism, which places no great

weight on the distinction between categorial and semantic intentions,

does not face such a problem. Yet moderate intentionalism is attuned to
other distinctions between intentions, beginning with the difference

between intentions that are never acted on (such as an artist's aban

This content downloaded from 59.189.72.200 on Tue, 31 Jan 2017 10:12:28 UTC
All use subject to http://about.jstor.org/terms

INTENTIONALISM IN AESTHETICS

doned musings about what she plans to do at some point in the future),

and those that are acted on and actually orient the artist's work in a
medium. In some cases, the action involved is a mental one, and what
the artist does is to make a decision, thereby constituting the work in a
particular manner.18 Such intentions cannot countervene facts about the

actual text or structure that has been produced, but when no such
conflict obtains, they can decisively inform both the artist's and critic's
thinking about the work. Such, at least, is the key claim of moderate

intentionalism.

V. A Test Case
As a result of his commitment to the uptake of the actual artist's
categorical intentions, Levinson's hypothetical intentionalism differs

from the kind of fictionalist intentionalism that I discussed earlier, but it

is not clear that his approach avoids all of its problems. We can see this
if we focus on a case where the hypothetical intentionalist disregards the

actual author's semantic intentions. Suppose we have a work of prose

fiction where the words and sentences in the text, standardly and
literally interpreted in a holistic way, are compatible with at least two
significant and incompatible interpretations, each of which appears to

provide an excellent, if not optimal, reading of the work. In one

reading, Rl, the governess who is the narrator of the work's embedded
tale is really quite mad; it is true in the story that she wrongly believes
and sincerely narrates that there are ghosts in the manor. In the other,
rival reading, R2, of which the text also allows, this narrator is distraught,

but not deluded; she detects and reports the presence of supernatural
beings, and it is true in the story that these malevolent entities exist.

Now, suppose as well that the author is known to have intended the
latter reading. He aimed at creating a ghost story in which the horrors
would be more terrible because presented only indirectly through the
report of an observer. As he put it, "prodigies, when they come straight,
come with an effect imperilled; they keep all their character, on the
other hand, by looming through some other history."19 That is why he

penned a text in which we only hear about the ghosts through the
account of the governess; the reading of that account is, in any case,
clearly framed as part of an exchange of ghost stories. And it just so
happens that the author's contemporaries, including such astute read
ers as Virginia Woolf, knew all this and never hesitated to call the work
a ghost story.20
With regard to such a case, moderate, actual intentionalism rules that

R2 alone is right; this is a ghost story, albeit one where the ghosts are

This content downloaded from 59.189.72.200 on Tue, 31 Jan 2017 10:12:28 UTC
All use subject to http://about.jstor.org/terms

841

842

NEW LITERARY HISTORY

only implicitly presented. This reading is supported by reference to the

text and to both the writer's semantic intentions and the related,

categorial intentions.
What does a hypothetical intentionalist say about such a case? A first
problem is that of sorting out the intentions. Is the author's intention to
make ghosts part of the story a purely semantic intention? What is its
relation to the categorial intention of writing a certain type of story,

namely, a ghost story? How could the author have framed the latter
intention without also having some sort of semantic intention relevant
to the presence of ghosts in the tale? It looks like a case of a single,

mixed intention, not two significantly distinct aims. In that case, an

interpreter working with hypothetical intentionalism has to decide
whether the semantic aspect disqualifies the intention. I have not been

able to find a principled basis for making such a decision (given, of

course, the text's compatibility with the relevant intentions).
Suppose the hypothetical intentionalist determines that the author's
intention to write a ghost story is a separate, purely categorial one. In
that case he or she should agree that R2 is correct. But let us stipulate
that the facts in the case are different, and that the author had only
semantic intentions about there being ghosts in the story (ghosts, that is,
that would loom through some other history). We stipulate, then, that
the author had no related, categorial intentions about the work being a

ghost story of some kind.21 The example is, we suppose, logically and
psychologically possible: let the author be a super-Crocean having the
deluded belief that his semantic and other artistic aims are unique. Now,
in that case, the interpreter who follows the strictures of hypothetical
intentionalism should not let the fact about the semantic intention tip

the scales in favor of R2. The work is, instead, seen to be ambiguous
between Rl and R2; its meaning is their exclusive disjunction.
At this point another feature of Levinson's hypothetical intentionalism

should be noted. The meanings discovered are now attributed to the

actual author?only hypothetically so. Being intended, the h?sitation has

a different status than it would have had had one decided that it was an
ambiguity resulting from authorial failure or ineptitude. The work is
better when we abandon the latter hypothesis in favor of the former, and
we are right to appreciate it as such.22 So although I know that the actual
author wrote the story with semantic intentions, I hypothesize?but do
not feign or imagine?an author who did not have those intentions, but

who intended for the work to be ambiguous. I do this because actual

semantic intentions are not constitutive, and because the work has
greater artistic value when I work with this hypothesis about the author's
intentions.

So in this instance hypothetical and moderate intentionalism do not

This content downloaded from 59.189.72.200 on Tue, 31 Jan 2017 10:12:28 UTC
All use subject to http://about.jstor.org/terms

INTENTIONALISM IN AESTHETICS

843

converge; instead, they present us with a choice between two rival

interpretive conclusions:
(MI) the author meant to imply the presence of ghosts in the story,

and since the text can be squared with that intention, the author
successfully did so, and R2 is correct;
(HI) the hypothetical actual author's intention is a story ambiguous
between the narrator's madness and sanity; the exclusive disjunction

(Rl or R2) and not (Rl and R2), is correct.

What reasons can we given to justify a choice between the principles

yielding these two options? Basically, the accounts differ with regard to
what they claim about the relation betweeen the work's causal history
and its meanings. Hypothetical intentionalism is selective about what
facts and evidence relevant to that causal history can be decisive in an
account of meaning. As the actual author's semantic intentions are not
decisive (given the text's ambiguity), the hypothetical actual author is

determined to be someone who intended to write an ambiguous tale.

Ambiguity cannot, it would seem, be an unintended feature of the work's
content.

Why should the possibility of recognizing unintended semantic ambi

guity matter? Why, more generally, should we care about the relation
between the work's causal history and its meanings? The answers rest on

assumptions that are, I think, shared by friends of hypothetical and

moderate intentionalism. Conclusions about meanings are often rele
vant to judgments concerning the artist's achievement. Someone who
tries to write a straightforward, unambiguous story, but ends up writing

something that everyone reads as involving a complex rhetoric of

unreliable narration, may have written something fascinating to read;
but this person's work should not be prized as the artistic achievement
of devising an unreliable narration. We want an interpretive theory that
is attuned to the difference between glorious serendipity and unfortu
nate failures, as well as the difference between the skillful realization of

valuable and difficult aims and the routine realization of lowly or

mediocre goals.23 Ambiguity, or lack thereof, is one such relevant aim.
Our appreciation of artists' achievements in this respect depends on our
working with an explanation of the work's genesis that is as accurate as

the evidence allows, an ideal to which the selective overruling of

semantic intentions is inimical. To echo Levinson's own phrases, artists
should not be allowed to make a work mean, by fiat, whatever they want
them to mean; yet critics should not, by selective weighing of evidence,
convert unintended meanings into intended ones, not even "by hypothesis,"

This content downloaded from 59.189.72.200 on Tue, 31 Jan 2017 10:12:28 UTC
All use subject to http://about.jstor.org/terms

844

NEW LITERARY HISTORY

because this sort of move obscures the historical role of semantic

intentions, including unsuccessful ones, in the making of the work.

Conclusion
The moderate intentionalist holds that the theory of appreciation and
interpretation should be attuned to the artist's constitutive role in the
making of works. It is the artist who, within "natural and logical limits,"
makes the work, and choosing and settling on categories and meanings

is part of that creative process. We ought to reject the criticism

promoting idea that it is the reader who invents the story; we prefer,

instead, a communicative model in which the reader attempts to

discover the nature of the story as told, acknowledging that it is the
storyteller who, within limits and contingent on his or her ability,
decides what happens in the story he or she is going to tell, including

events that need not, for various reasons, be related directly in the text.
Hypothetical intentionalism suffers from the problem that we do not

have any systematic way to separate the categorial wheat from the
semantic chaff. What is more, it is not even obvious that we have any

good grounds for trying to do so. Some intentions are inextricably

semantic and categorial; some chaff is categorial, and there is semantic

wheat to be harvested. If the works of art that actual authors have

created are the prime target of an interpretive hypothesis, then we
should let all of the available evidence about the causal history of the
artistic structure have the same, initial status. Part of that history is a
matter of the semantic intentions on which the artist has successfully or
unsuccessfully acted.24 Sometimes the author's semantic intentions are
less limited than the meanings a reader may be able to dream up on the
basis of the text and other background evidence. Sometimes interviews
and diaries open up all sorts of wonderful undiscovered meanings. We

can indeed imagine a Kafka whose diary reveals stupid semantic

intentions, but we can also actually read the remarkable diaries of the

real Franz Kafka. What is harder to imagine is why critics should be

required to refrain from allowing their interpretations of Kafka's works
to be in any way guided by an interpretation of these fascinating diaries

and other evidence relevant to the actual author's thoughts and

experience. Recognizing that in some cases limited or boring semantic

intentions are decisive of a work's features is the price we pay for an

interpretive principle that allows us, on other, happier occasions, to

recognize that the artist's laudable and complex aims were decisive.

?rhus University

This content downloaded from 59.189.72.200 on Tue, 31 Jan 2017 10:12:28 UTC
All use subject to http://about.jstor.org/terms

INTENTIONALISM IN AESTHETICS 845

NOTES
1 For background, see Intention and Interpretation, ed. Gary Iseminger (Philadelphia,
1992); Understanding the Arts, ed. Jeanette Emt and G?ran Hermer?n (Lund, 1992); and
Berys Gant, "Interpreting the Arts: The Patchwork Theory," The Journal of Aesthetics and Art

Criticism, 51 (1993), 597-610.

2 For more background on intentions, see Paisley Livingston and Alfred R. Mele,
"Intention and Literature," Stanford French Review, 16 (1992), 173-96; and Alfred R. Mele,
Springs of Action (New York, 1992).
3 For valuable background, see Mette Hjort, The Strategy of Letters (Cambridge, Mass.,
1993), and Lynne Rudder Baker, Explaining Attitudes: A Practical Approach to the Mind

(Cambridge, 1995).
4 See George M. Wilson, "Again, Theory: On Speaker's Meaning, Linguistic Meaning,

and the Meaning of a Text," in Rules and Conventions: Literature, Philosophy, Social Theory, ed.

Mette Hjort (Baltimore, 1992), pp. 1-31.

5 The work by Jorge Luis Borges is translated by Anthony Bonner as "Pierre Menard,
Author of Don Quixote," in Ficciones (New York, 1962), pp. 45-55. For some of the
philosophical extrapolations, see Arthur Danto, The Transfiguration of the Commonplace
(Cambridge, Mass., 1981); David Davies, "Text, Context, and Character: Goodman on the
Literary Artwork," Canadian Journal of Philosophy, 21 (1991), 331-45; Gregory Currie,
"Work and Text," Mind, 100 (1991), 325-40; andjerrold Levinson, "What a Musical Work
Is," in his Music, Art, and Metaphysics (Ithaca, 1990), pp. 63-88. A similar point was made
much earlier by Carl Lange in a little-known treatise on aesthetics, Bidrag til Nydelsemes
Fysiologie som Grundlag foren rationel JEstetik (Copenhagen, 1899), p. 125. Lange argued that

two physically identical paintings could have very different artistic values. The one
produced first has the virtue of novelty and can thereby offer the special pleasure of
admiration made possible by our awareness of this feature of the work, which is lacking in

any subsequent copies.

6 No?l Carroll, "Art, Intention, and Conversation," in Intention & Interpretation, pp. 97

131.

7 George D. Painter, Marcel Proust (Harmondsworth, 1982); Vincent Descombes, Proust:

Philosophie du roman (Paris, 1987); Ren? Girard, Mensonge romantique et v?rit? romanesque

(Paris, 1961).

8 Alexander Nehamas, "Writer, Text, Work, Author," in Literature and the Question of
Philosophy, ed. Anthony J. Cascardi (Baltimore, 1987), pp. 267-91, and "The Postulated
Author: Critical Monism as a Regulative Ideal," Critical Inquiry, 8 (1981), 131-49. For
insightful criticisms, see Robert Stecker, "Apparent, Implied, and Postulated Authors,"
Philosophy and Literature, 11 (1987), 258-71.
9 The importance of multiple authorship in literary history is ably demonstrated by Jack
Stillinger in his Multiple Authorship and the Myth of Solitary Genius (New York, 1991);
Stillinger does not, however, discuss cases where several persons successfully engage in
collective authorship following a shared plan or intention.
10 I say "broadly" because I intend to remain neutral on many of the controversies in
contemporary pragmatics. For background, see Paul Grice, Studies in the Way of Words
(Cambridge, Mass., 1989); Stephen C. Levinson, Pragmatics (Cambridge, 1983), pp. 97
166; and Fran?ois Recanati, Meaning and Force: The Pragmatics of Performative Utterances
(Cambridge, 1987), pp. 118-21.
11 See Jerrold Levinson, "Intention and Interpretation in Literature," in his The Pleasures
of Aesthetics: Philosophical Essays (Ithaca, 1996), pp. 175-213, hereafter cited in text as I; for

the discussion of the Kafka example, see pp. 184-86. A slightly different version of this
essay appeared as "Intention and Interpretation: A Last Look," in Intention ?f Interpretation.

This content downloaded from 59.189.72.200 on Tue, 31 Jan 2017 10:12:28 UTC
All use subject to http://about.jstor.org/terms

846 NEW LITERARY HISTORY

12 William E. Tolhurst, "On What a Text Is and How It Means," British Journal of Aesthetics,

19 (1979), 3-14.

13 See Jerrold Levinson, "Extending Art Historically," in The Pleasures of Aesthetics, pp.

150-71; the example is given on p. 158, n. 8.

14 For background, see Alfred R. Mele and Paul K. Moser, "Intentional Action," Nous, 28

(1994), 39-68.

15 See Levinson, "Extending Art Historically," p. 158, n. 7 for this sort of emphasis on
first-order and second-order intentions.

16 I have in mind the account proposed by Gregory Currie in The Nature of Fiction
(Cambridge, 1990), ch. 1. It stands to reason that if making fiction is communicatively
intending that something be imagined, the intended something has to be the content of
a semantic intention. "Semantic," after all, is not the sharpest of terms, and the best we
have is the old "relations between signs and that for which they stand" idea that comes to
us from Charles Sanders Peirce via Charles Morris. Attempts to give a sharper demarcation
of semantic/nonsemantic in terms of such notions as meaning and reference only shift the
burden onto another pair of frail shoulders.
17 An example is the so-called first trilogy by the Japanese writer Natsume Soseki. For

background, see Norma Moore Field, "Afterword," And Then: Natsume Sosekis Novel
Sorekara (Rowland, Vt, 1988), pp. 258-78.
18 For example, the artist's second-order decision that her own creative work on a given
piece is completed is crucial to the commonly applied distinction between unfinished and
finished works of art, where the latter category includes such purposefully "incomplete"

items as romantic fragments and ruins. For more on this topic, see my "Counting
Fragments, and Frenhofer's Paradox" (forthcoming in British Journal ofAesthetics).
19 Henry James, "From a Preface," in The Turn of the Screw, ed. Robert Kimbrough (New

York, 1966), p. 103.

20 Virginia Woolf, "Henry James's Ghosts," in The Turn of the Screw, ed. Kimbrough, pp.

179-80.

21 Here we probably depart significantly from the actual case of Henry James's writing of
The Turn of the Screw. As my aim in evoking the example is primarily a matter of conceptual

clarification, this does not matter to the argument.

22 Note, however, that this is not a reading whereby the governess is known to be an
unreliable narrator, which would require that, in spite of some misleading evidence to the
contrary, we know that she is deluded. For a valuable clarification of concepts of ambiguity
and unreliability, see Gregory Currie, Image and Mind: Film, Philosophy, and Cognitive Science

(Cambridge, 1996), ch. 9. Currie would say that the intentions in question are those of the
implied author (an interpreter's construct), not the actual author.
23 To risk an analogy, in some games we keep score by judging whether the shot made
was the shot called; in others, the scoring system does not filter out lucky shots. The
appreciation of art is more like the former than the latter. One reason why categorial
intentions are always relevant, and sometimes decisive, to interpretation is that artistic
appreciation is attuned to the relation between aims and achievements. My claim is that
the same reasoning holds with regard to semantic intentions.
24 Thanks to Jerry Levinson, David Davies, and Al Mele for helpful comments on a draft
of this paper, a version of which was presented at the Nordic Society for Aesthetics meeting

in Oslo in May 1997. Financial support for this research was provided by the Social
Sciences and Humanities Research Council of Canada and by the Fonds pour la Formation
des Chercheurs et VAide ? la Recherche of Quebec.

This content downloaded from 59.189.72.200 on Tue, 31 Jan 2017 10:12:28 UTC
All use subject to http://about.jstor.org/terms