Art and Understanding
In Defence of Aesthetic Cognitivism
We praise certain artworks for their profundity and subtlety, for the insights they provide or
for how they make us see the world anew and we think these features are artistically relevant.
We criticize other works for their shallowness, superficiality or sentimentality and think them
thus artistically flawed. These are artistic evaluations that seem also to be, or to depend on,
cognitive evaluations. Aesthetic cognitivism takes such features of our evaluations of art-
works seriously. It is best thought of as a conjunction of an epistemic and an aesthetic claim:1
(1) Epistemic claim: Artworks have cognitive functions.
(2) Aesthetic claim: Cognitive functions of artworks partly determine their artistic value.
Aesthetic non-cognitivism is the denial of either or both of the epistemic and aesthetic claims.
This kind of aesthetic cognitivism must be distinguished from cognitivism regarding aesthetic
judgments; a position corresponding to the kind of cognitivism discussed in meta-ethics and
usually defined by the thesis that aesthetic judgments can be true or false.
It is important not to construe the epistemic and the aesthetic thesis as too strong. The
epistemic thesis (1) does not claim that all artworks have cognitive functions. I presume that
works of all of the arts but not all works of an art have cognitive functions; hence, that (1)
holds true not only for the representational arts, such as literature, film and figurative paint-
ing, but also for abstract painting, pure music, non-narrative dance, and even for architecture.
However, I will not argue for that presumption here, and most of my examples will be taken
from the representational arts. The aesthetic thesis (2) does not claim that all cognitive func-
tions of artworks increase their aesthetic value. We can use artworks as sources of infor-
mation in a manner that has nothing to do with them being artworks. Often, an artwork will
teach us something about its artist, and ancient artworks reveal much about the cultures that
produced them. But such a function of artworks as biographical or historical evidence does
not reveal that they are therefore better works of art. Hence, only those cognitive functions an
artwork has as an artwork, thus belonging to its proper functions, are artistically relevant.
Furthermore, claim (1) leaves room for other functions, such as practical, decorative, political
and economic ones, and can acknowledge that the importance of cognitive functions differs
across the arts (and within an art for different works). While they may be central to the repre-
sentational arts, they might be less eminent in abstract painting, pure music, non-narrative
dance and architecture. Claim (2) leaves room for a plurality of artistic values, of which cog-
nitive ones are just one type. Cognitive functions of artworks only partly determine their
In this paper, my first aim is to defend the epistemic thesis of aesthetic cognitivism. Since
it seems undeniable that artworks have cognitive functions, yet less clear whether they have
1 Gaut 2005, 436; 2006, 115; 2007, 137.
Final draft (March 2011). Forthcoming in Christoph Wagner; Marc Greenlee;
Rainer Hammwöhner; Bernd Körber; Christian Wolff (ed.): Bilder sehen.
Perspektiven der Bildwissenschaft. Regensburg: Schnell + Steiner
(Regensburger Studien zur Kunstgeschichte).
them as artworks, I will focus on cognitive functions that plausibly belong to the proper func-
tions of artworks. My second aim is to sketch a suitable epistemological framework for aes-
thetic cognitivism. This is important since, as David Novitz remarked, even if many philoso-
phers have written about art and cognition, »what they have not done is examine the episte-
mological underpinnings of their various claims«.2 I will try to achieve my first aim by realiz-
ing my second aim and hence argue for the thesis that artworks have cognitive functions by
sketching a suitable epistemological framework. In Section 1, I present reasons for conceiving
of epistemology as a theory of understanding rather than a theory of knowledge. Art will enter
the stage only in Section 2, where I will show that an epistemology of understanding can and
should accommodate the cognitive functions aesthetic cognitivists claimed for artworks,
hence providing a suitable epistemological framework. To argue in this way in favour of the
claim that artworks have cognitive functions does not beg the question since my argument for
an epistemology of understanding does not depend on my aim to defend the epistemic thesis
of aesthetic cognitivism. In Section 3, I will answer some of the most pressing non-cognitivist
objections within the proposed framework.
1. From Knowledge to Understanding
Epistemology is usually conceived of as the theory of knowledge concerning the nature,
sources and limits of knowledge.3 Knowledge is identified with propositional knowledge and
analysed in terms of justified (or reliably generated) true belief, perhaps extended by a further
condition designed to avoid Gettier counter examples. Cognitive advancement is construed as
the growth of knowledge. It is accomplished by the acquisition of new justified (or reliably
generated) true beliefs that satisfy the additional condition.
In recent years, several epistemologists have suggested to take understanding rather than
propositional knowledge as the central concept of epistemology. Nelson Goodman and Cathe-
rine Elgin argue that the main goal of our cognitive endeavours is not to acquire justified (or
reliably generated) true beliefs but to advance our understanding.4 The proposed revision does
not suggest giving up the concept of propositional knowledge and replacing it by a concept of
understanding. As we will see, having propositional knowledge is an important part of under-
standing a phenomenon. Hence, an epistemology of understanding has to comprise an account
of propositional knowledge.
1.1 Division of Understanding
The term »understanding« is used in a variety of ways. According to the use that is of primary
interest for epistemology, it designates a cognitive success or achievement. From an epistemic
point of view, two grammatical forms involving understanding stand out. In case of what Jon-
athan Kvanvig calls »objectual understanding«,5 the grammatical form takes an object; as
when we say that Sophie understands the history of the Soviet Union or that Paul understands
thermodynamics. In case of propositional understanding, the grammatical form employs a
2 Novitz 2004, 993.
3 See e.g. Klein 2005.
4 Goodman/Elgin 1988; Elgin 1996; 2006. My argument is primarily based on their work. More recently, vir-
tue epistemologists have proposed to define intellectual virtues as capacities or character traits that contribute
to achieving the epistemic goal of understanding (Riggs 2003; Roberts/Wood 2007; Zagzebski 2001). Fur-
thermore, it has been argued that referring to understanding may avoid the value problem for knowledge
(Kvanvig 2003; Pritchard 2010).
5 Kvanvig 2003, 191.
›that‹-clause; as when we say that Sophie understands that the Soviet Union collapsed or that
Paul understands that heat cannot spontaneously flow from a colder location to a hotter loca-
tion. There is also a range of attributions of understanding followed by ›wh‹-clauses: Sophie
understands why the Soviet Union collapsed, where Vladimir comes from, what it takes to be
a student and when it is time to go. But such uses can presumably be either explained in terms
of propositional or in terms of objectual understanding.6 Objectual understanding is the core
conception of understanding. Understanding is primarily related to a fairly comprehensive
body of information. The understanding expressed in individual propositions derives from an
understanding related to larger bodies of information that include those propositions. Sophie
understands that the Soviet Union collapsed, because she grasps how the proposition stating
that fact fits into and is justified by reference to a more comprehensive understanding of the
history of the Soviet Union.7 In what follows, »understanding« is used for objectual under-
standing and »knowledge« for propositional knowledge.
The mentioned cases of what could be called »factual understanding« must be distin-
guished from semantic or, more generally, symbolic understanding, like for example when we
say that Sophie understands the sentence »The Soviet Union collapsed« or that Peter under-
stands the Hertzsprung-Russell diagram; but also when we say that Paul understands the the-
ory of thermodynamics. Symbolic understanding is a genuine cognitive achievement. But one
can understand a sentence, a diagram or a theory without committing oneself to them and re-
gardless of whether the sentence is true, the diagram apt and the theory answers to the facts.
Neither of this holds for an understanding of what the sentence, the diagram and the theory
are about. Understanding a topic in terms of a theory, for instance, presupposes that the theory
is largely correct and that one commits oneself to it. Even if symbolic understanding is often a
precondition of factual understanding, I am primarily concerned with the second.8 As a conse-
quence, rather than with understanding artworks I will primarily be concerned with their con-
tributions to understanding aspects external to the works.
The use of »understanding« for a cognitive achievement is not the only epistemologically
relevant use of the term. We can distinguish between understanding as a cognitive faculty in
an inclusive sense consisting of a collection of abilities, understanding as the process of using
such abilities in our inquiries, and understanding as what the cognitive process achieves.9
Even if an epistemology of understanding focuses on cognitive achievements, it has to deal
with abilities and processes as well. Processes may play a crucial role in explaining what is
achieved by them, and the achievement can involve the improvement of abilities that are part
of understanding as a cognitive faculty.
1.2 Reasons for Revising Epistemology
Why should we take understanding rather than knowledge as the central concept of episte-
mology? The answer is that it more aptly captures what we consider a cognitive achievement
in science, philosophy and everyday life. Some knowledge is no cognitive achievement; some
cognitive achievements do not constitute knowledge, others go beyond it.
6 The first is proposed by Kvanvig 2003, 189.
7 Elgin 2009a, 322–323; Kvanvig 2003, 192.
8 The distinction between factual and symbolic understanding is not a sharp one. The reason relates to the
deficiencies of the analytic-synthetic distinction and to the holistic character of understanding (cf. Cooper
1994, 2; 1995, 206–207.
9 Goodman/Elgin 1988, 161–162.
Knowledge that is No Achievement
Knowledge of trivial or of irrelevant truths neither constitutes a cognitive achievement nor
contributes to understanding a given phenomenon. If the question is why the Soviet Union
collapsed, then neither to know that Moscow is a town nor to know that heat cannot sponta-
neously flow from a colder location to a hotter location advances our understanding of the
phenomenon in question. The first truth is trivial and knowledge of the second irrelevant for
answering the question at hand. But if cognitive advancement is simply construed as growth
of knowledge, we should concentrate on trivial or obvious statements like »Moscow is a
town«, irrespective of whether they are relevant for our question, since they are more likely
than intriguing ones to be true and justifiable.10
Achievements that Do Not Constitute Knowledge
Some cognitive achievements contribute to, and are part of, our understanding of a phenome-
non but do not constitute knowledge. Firstly, cognitive progress may be made by developing
categories that impose an order on a domain appropriate to our cognitive goals.11 Such a new
categorization advances our understanding by reorganizing a domain and thereby revealing
significant but hitherto overlooked or underemphasized likenesses and differences. Botany,
for example, reorganizes (and enlarges) the domain of fruits when it replaces the everyday
category of a berry as a small roundish juicy fruit without a stone by the category of a berry as
an indehiscent fruit with fleshy pericarp. Blackberries, raspberries and strawberries are berries
in the everyday sense but not in the biological sense; bananas, melons, tomatoes and zucchinis
are berries in the biological sense but not in the everyday sense. The reorganization reveals
that things like blueberries and strawberries that are superficially alike are deeply different
and things like blueberries and zucchinis that are superficially different are deeply alike. Such
a categorization contributes to, and is an essential part of, our understanding. But being non-
propositional, categories are neither true nor false and hence not the right sort of matter for
constituting or expressing knowledge. Of course, reorganizing a domain by an apt scheme of
categories leads to new knowledge, since reclassifying the objects of a domain involves ap-
plying categories to objects they had not previously been applied to, and hence to form new
propositions. But this is not the only cognitive contribution of a categorization since it can
promote other cognitive goals than truth, such as simplicity. Botany could have retained the
commonsense categorization and form the hypothesis that some, but not all, berries and some,
but not all, vegetables are indehiscent fruits with a fleshy pericarp.12 But the proposed cate-
gorization enables much simpler generalizations. The simplicity of an account can be an
epistemic desideratum. But it is no indication of its truth. Rather, the simplest account com-
patible with the evidence is typically less likely to be true than some of its rivals.
Secondly, cognitive progress may be made by developing adequate three-dimensional
models, drawing accurate maps and producing apt diagrams.13 By highlighting certain fea-
tures of objects and neglecting others, models, maps and diagrams provide perspectives on
them: ways of conceiving of the objects that enhance our understanding of them. A Corey-
Pauling-Kortum (CPK) model composed of closely packed spheres made of plastic highlights
structural relations it shares with the DNA molecules and neglects features like their size, col-
10 See Goodman/Elgin 1988, 135–152.
11 Elgin 1996, 104–105; 2006, 203–204.
12 The hypothesis only makes sense if the expression »fruit« is taken to be the botanic category which includes
vegetables, pods, nuts, ears and cones in the culinary sense.
13 Zagzebski 2001, 241.
our and material. The neglect is a merit rather than a defect. Being larger, colour-coded and
durable, the model makes the structural features manifest so that they can be discerned more
easily than by observing DNA molecules directly. The CPK model thereby provides a per-
spective on them: we can understand their helical structure in terms of it. In a comparable
way, we can understand the geology of a massif in terms of a geologic map and the evolution
of stars in terms of a Hertzsprung-Russell diagram. Hence, our understanding of a domain can
be couched in and conveyed by models, maps and diagrams. But being non-verbal symbols
and non-propositional in form, they are not the sort of things that can be believed and consti-
tute or express knowledge. Of course, they also usually lead to new knowledge. But this does
not make them redundant. Either the information they convey is not fully expressible in
propositional form or the inconvenience of the resulting descriptions would make them poor
substitutes for the non-verbal symbols.
Thirdly, cognitive progress may be made by asking new questions, clarifying the ques-
tions we are trying to answer or replacing misguided questions by better ones. Insights into
specific genes related to cognitive abilities made it possible to ask more refined and powerful
questions about development, heterogeneity, co-morbidity and gene-environment interplay.14
The ability to ask these questions is a cognitive achievement even if the behavioural sciences
are not yet in a position to answer them conclusively. Furthermore, according to G. E. Moore,
the difficulties and disagreements that have dogged philosophy are mainly due »to the attempt
to answer questions, without first discovering what question it is which you desire to
answer«.15 In his Principia Ethica, Moore distinguished clearly two kinds of questions which
philosophers »have almost always confused with one another and with other questions«16 –
What kind of things ought to exist for their own sake? What kind of actions ought we to per-
form? – and clarifies their meaning. This is a cognitive achievement independent of Moore’s
particular answers to these questions. It consists of a better understanding of the problems at
hand as well as the options for dealing with them. Finally, Goodman suggested replacing the
question »What is art?« by the question »When is art?«. Rather than asking what essential
features an object must have to be an artwork, Goodman urged us to ask under which condi-
tions an object functions as an artwork. He thereby shifts our attention from the question of
whether an object or event is an artwork to the question of how it functions.17 This shift has
advanced our understanding by allowing fizzled out debates to rest and by revealing new ave-
nues of inquiry worthy of exploration. All three examples suggest that understanding can be
located in insightful questions. But questions do not constitute knowledge since the proposi-
tions they express are asked rather than believed. Of course, questions also often lead to new
knowledge. Answers to them can often be expressed in that-clauses; if we believe them and
the beliefs are justified and true, they can constitute knowledge. But asking insightful ques-
tions in itself seems to be a considerable cognitive achievement; especially in philosophy,
where progress »often amounts to the clarification rather than the solution of problems«.18
Fourthly, cognitive progress may be made by becoming directly acquainted with sensory
or emotional qualities. Such direct acquaintance can deepen our understanding by providing
knowledge of how it is like to have a certain experience or emotion, or to be in a certain situ-
ation. Knowing how it is like to taste coffee can deepen our understanding of coffee, knowing
14 Plomin/Craig 2001, 47.
15 Moore 1903, vi.
16 Moore 1903, vi.
17 Goodman 1978, 57–70.
18 Glock 2008, 106.
how it feels like to be bereaved our understanding of bereavement, knowing how it is like to
live as an illegal immigrant in the United States our understanding of the problem of illegal
immigrants. But this sort of knowledge is irreducibly non-propositional in the sense that it
cannot be adequately captured by linguistic descriptions. No matter how precise and vivid
your descriptions are, they will never capture the whole content of my acquaintance with the
taste of coffee or the feelings of an orphan.19 Nonetheless, direct acquaintance often provides
us with new justified beliefs, or it improves the justification of beliefs we already have. My
acquaintance with the taste of a certain coffee may result in a new justified belief about its
quality or confirm such a belief if I held it already. But the epistemic value of direct acquaint-
ance is not exhausted in its providing new beliefs or forming grounds for them.20 Even if by
visiting illegal immigrants in the United States I were to gain no additional justified (or relia-
bly generated) beliefs about their living conditions, nor improve the justification for the ones I
have, I might still deepen my understanding by acquiring a sense for what it is like to live as
illegal immigrant in the United States.
Finally, in science as well as in philosophy, cognitive progress is often made by contriving
idealizations and thought experiments. They are critical to or even constitutive of the under-
standing that science and philosophy deliver. But they do not constitute knowledge since they
are not true and do not even purport to be true.21 Idealizations characterize ideal cases that do
not and perhaps cannot occur. Nothing in the world exactly answers to them; hence, as de-
scriptions they are false. Nonetheless, they contribute to scientific understanding. Idealiza-
tions are fictions designed to afford epistemic access to matters of fact that are otherwise dif-
ficult or impossible to discern. They do so by highlighting features that are instantiated in real
cases or diverge at most negligibly from them. They thereby enable us to explore these fea-
tures and their causes and consequences by disregarding complications that overshadow them
in real cases. The ideal gas law, for example, accounts for the behaviour of actual gases by
describing the behaviour of a gas composed of perfectly elastic, spherical molecules that
occupy negligible space and exhibit no intermolecular attraction. There is no and cannot be
such a gas. Nonetheless, scientists understand the behaviour of actual gases by reference to
the ideal gas law. In circumstances where divergence from the ideal is negligible (roughly, in
cases of monatomic gases at high temperature and low pressure), the ideal gas law, although
not strictly true of actual gases, is true enough of them. Where the idealization is true enough,
it is illuminating to think of actual gases as displaying the interdependence of temperature,
pressure, and volume that is highlighted by the idealization. Whether an idealization is true
enough is a contextual matter and depends on the degree of precision we want or need. Where
the divergence from the ideal is not negligible, corrections have to be introduced. Instead of a
simple true description of the behaviour of an actual gas, we get a complicated truth that
makes reference to deviations from the ideal. This might seem to suggest that the real cogni-
tive contribution resides not in the idealization but in the knowledge it leads to. But since
truth is not our only cognitive goal and can sometimes be overridden by others like simplicity
and applicability, an idealization may at least in certain contexts be preferable to the truth it
19 This is sometimes taken as a reason for claiming that one can come to know how it is like to taste coffee only
by tasting it, or that one does not know what it feels like to be bereaved until one has experienced bereave-
ment (cf. Novitz 1987, 120).
20 Roberts/Wood 2007, 33.
21 My characterization of how idealizations and thought experiments function follows Elgin. But it does not
commit me to Elgin’s way of spelling it out in terms of Goodman’s notion of exemplification; see Elgin
1996, 180–196; 2006, 210–213; 2009a, 326–329; 2004, 122–128; 2005, 47–48; 2009b, 77–90.
approximates. The ideal gas law may advance our understanding of certain gases in ways the
unmanageably complicated truth would not.
Thought experiments are fictions contrived to reveal what would happen if certain condi-
tions were met. They are not actual, and often not even possible, experiments. The imaginary
conditions that set their stage do not obtain and are often not even physically possible. None-
theless, if their assumptions about what can be fruitfully neglected are correct, thought
experiments can afford an understanding of the phenomena they pertain to. They highlight
certain features, display their significance in the imaginary setting, thereby giving reasons to
suspect that these features are also salient in related real situations. They draw out implica-
tions of these features and thereby revealing unrecognised or unappreciated commitments.
Those commitments may speak in favour of or against the hypothesis under investigation. By
considering a famous violinist being in coma who has been hooked up to you while you were
asleep and who is only able to survive if he remains so for nine months, Judith Thomson sup-
ports the claim that abortion could be morally permissible even when the foetus has a right to
life.22 The fact that no violinist has ever been and probably no one will ever be hooked up to
another person does not discredit Thomson’s thought experiment. Thought experiments used
negatively to undermine a hypothesis are often at the same time used positively to establish an
alternative hypothesis. By considering how a light body tethered to a heavy body would fall,
Galileo refutes the Aristotelian hypothesis that heavier bodies fall faster than light ones and
establishes that the rate at which objects in a vacuum fall is independent of their weight.23
Sometimes, the commitments revealed by a thought experiment simply illustrate a hypothesis,
making it clear and evident; or they flesh it out, enriching our understanding of what its ac-
ceptance would commit us to. By considering a cannon shooting a cannon ball further and
further until the earth curves away as fast as the ball falls, with the eventual result being that
the cannon ball will return to the spot where it was fired and go around again and again,
Newton illustrated how the moon is kept in its orbit in just the same way as an object falls to
the earth.24 By considering the fate of an imaginary cat penned up in a steel chamber, Schrö-
dinger drew out bewildering implications of quantum mechanics that are in conflict with
some powerful common sense beliefs about macro-size objects such as cats.25
The cognitive achievements listed in this Subsection are part of our understanding but do
not constitute knowledge. In contrast to knowledge, understanding need not be couched in
true beliefs or statements. It might equally be located in apt verbal categories, effective non-
verbal symbols, insightful questions, direct acquaintance and revealing fictions.
Achievements that Go Beyond Knowledge
Some cognitive achievements go beyond knowledge but are involved in our understanding of
a domain. Striving for understanding is considerably more ambitious than acquiring
knowledge. Besides knowing the important and relevant truths that belong to a comprehen-
sive, coherent account of a domain and comprehending the appropriate fictions (like idealiza-
tions and thought experiments), understanding comprises grasping how the various truths and
fictions relate to each other and to further elements of the account (like categories, non-verbal
symbols and questions), and being able to use the information: to argue within the framework
of the account, to apply its results to new situations, to assess and acknowledge its limits, to
22 Thomson 1971, 47–66.
23 See Palmieri 2005, 223–240.
24 See Ducheyne 2006, 435–437.
25 Brown/Fehige 2010.
devise suitable (thought) experiments, to ask new questions unto which the account does not
yet provide conclusive answers, and so on.26 Understanding thermodynamics, for example,
involves not only knowing the important thermodynamic truths and comprehending the rele-
vant idealizations; it also involves grasping coherence-making relationships in the large and
comprehensive body of information provided by thermodynamics; and being able to provide
thermodynamic explanations of new cases, to assess their limits for a given phenomenon, to
design and execute thermodynamic experiments, to draw out implications of thermodynamic
findings, to answer a variety of questions and ask new questions unto which thermodynamics
does not yet provide an answer, and so on. In contrast to physical understanding, historical
understanding of something like the collapse of the Soviet Union concerns one particular
event. But this too does not amount to nothing more than knowing certain truths. Moreover, it
involves grasping connections between them and further elements as well as being able to use
such information; even if the application of its insights are more restricted and more tentative.
1.3 Characteristics of Understanding
As a result of the above discussion, a number of characteristics emerge distinguishing under-
standing from knowledge. Firstly, understanding is not a species of belief. Its content cannot
even be fully explicated as a collection of beliefs since it involves grasping connections be-
tween beliefs, non-belief states like questions, non-propositional commitments like categories
and non-verbal symbols, as well as having certain cognitive abilities.
Secondly, understanding is holistic. Knowledge can be broken down into discrete bits. It
is knowledge of an individual fact, expressed by a proposition. The proposition is true, the
knower believes the proposition and his belief is justified (or reliably generated). If the con-
tent or the justification of a belief is taken to depend on relations it bears to experiences and
other beliefs of the knower, this may introduce a holistic element. But even if knowledge is
partly holistic, understanding is wholly holistic.27 It cannot be broken down into discrete bits.
It is the understanding of a whole domain or topic, expressed in a more or less complex ac-
count or theory containing propositional and non-propositional elements. The account or the-
ory answers to the facts and the understander is committed to it and justified in it.
Thirdly, unlike knowledge, understanding is gradual. For any fact, either one knows it or
one does not know it. But understanding admits of degrees. Sophie has some understanding of
the history of the Soviet Union, while her tutor has a greater understanding and her professor
for the history of Eastern Europe an even greater understanding. Understanding can vary at
least in breath, depth, significance and accuracy.28 The professor’s understanding is broader
than Sophie’s understanding since it is embedded into a more comprehensive understanding
of Eastern European history. It is deeper since the web of his commitments is more tightly
woven. It is more significant since he weights some facts they both recognize more appropri-
ately. And it is more accurate since most of his beliefs are true or at least close to the truth
while some of Sophie’s beliefs are still more or less the crude characterizations a novice starts
26 Elgin 2009a, 323. While the ability to use information is seldom mentioned, grasping of interconnections is
often taken to be the outstanding feature of understanding (cf. Kvanvig 2003, 192; Riggs 2003, 217; Rob-
erts/Wood 2007, 47, 56; Zagzebski 2001, 241, 244).
27 A holist about belief or about justification will even construe knowledge wholly holistic. But unlike under-
standing, knowledge need not be construed wholly holistic; and at least for certain kinds of knowledge, such
as observational knowledge, a wholly holistic construal does not seem to be very plausible.
28 Elgin 2009a, 324–326.
Fourthly, in contrast to knowledge, understanding is not factive.29 One can only know that
p if »p« is true. But one’s understanding can involve propositions that are not true; and some
of them may even belong to the central propositions that constitute the account of the topic.
As we have seen in the preceding paragraph, understanding can be more or less accurate.
Novices like Sophie as well as scientists start out with crude characterizations that properly
direct them towards their topic and then refine these characterizations. Their advancement of
understanding involves a move from beliefs that are strictly false but in the right neighbour-
hood to beliefs that are closer to the truth. The development may result in true beliefs. But
even an earlier step displays some measure of understanding. Otherwise we would have to
deny that science yields any understanding, since scientific theories do not largely consist of
truths with a few relatively insignificant falsehoods at the periphery. Furthermore, as I have
argued in the preceding Section, even mature science is full of idealizations and thought ex-
periments. In contrast to the more or less crude characterizations, they are not in need of im-
provement and not supposed to be eliminable from scientific theories. Neither can they be
banished to the periphery of theories. Even if understanding is not factive, it must, of course,
answer to the facts by accommodating the evidence. But since understanding is holistic, ac-
commodating the evidence is a requirement on the entire theory, not on each individual ele-
ment of it. Hence, understanding a topic does not imply that all central beliefs that constitute
our theory are true.
Finally, understanding is related to a plurality of epistemic goals. Knowledge admits of an
account that takes truth to be the only or at least the highest epistemic goal. If such an epis-
temic value monism acknowledges other epistemic goals besides truth, then it claims that they
are goals only insofar as they facilitate our getting to the truth. Understanding, in contrast,
demands epistemic value pluralism. Categories, non-verbal symbols and questions are neither
true nor false; idealizations and thought experiments are known to be false but nonetheless
epistemically valuable. According to epistemic value pluralism, truth is only one of many
epistemic goals and can be overridden by others, such as generality, simplicity, parsimony,
robustness, explanatory power and applicability.
2. Art’s Contribution to Understanding
Having argued for conceiving of epistemology as a theory of understanding rather than
knowledge, I will argue that an epistemology of understanding can and should accommodate
the cognitive functions aesthetic cognitivists have claimed for artworks. Most cognitivists
hold that artworks can provide knowledge. But they often insist that this is a minor and com-
paratively insignificant part of what we learn from artworks.30 I will return to the function of
providing knowledge in Section 3. In this Section, I focus on cognitive contributions that
either are not considered knowledge or go beyond it, following the structure of my argument
in Section 1. The proposed list is neither exhaustive nor exclusive. Artworks may make fur-
ther contributions to understanding and usually contribute in more than one way. Let me start
with contributions that do not constitute knowledge.
29 Elgin 2009a. Kvanvig (2003, 190–191), in contrast, construes objectual understanding as factive. Other au-
thors who take understanding to be factive focus exclusively on understanding why (e.g. Pritchard 2010, 75–
30 See Novitz 1987, 133; Gaut 2005, 439.
2.1 Contributions that Do not Constitute Knowledge
Artworks, especially literary works, can provide us with new categories for classifying actual
objects.31 This is even the case with fictional works. Fictional terms do not denote when taken
literally; but they can metaphorically apply to actual objects and persons. A man who devotes
himself to a preposterous and hopeless but nonetheless noble undertaking is metaphorically a
Don Quixote; a man who seduces women with ease and lets them down soon after is meta-
phorically a Don Juan. Both fictional singular terms can be used metaphorically as predicates
for actual people. In grasping the metaphors, one engages in an open-ended exploration of the
salient similarities between the fictional character and some actual person. Other terms are
introduced as metaphors in the works themselves. In On Love, Stendhal used the term »crys-
tallization« that literally denotes a physical process as a metaphor for the mental process in
which unattractive characteristics of a new love are transformed into perfections. According
to this metaphor, nascent interest in a loved person conceals her (or his) real features by flat-
tering illusions analogous to the little salt prisms hiding a leafless branch of hornbeam and
glittering like the finest diamonds.
Furthermore, literary works often provide us with neologism. Some of them are derived
from well-known characters (»quixotic«) or from famous authors (»Orwellian«). Others have
been coined by the authors. In Lolita, Nabokov introduced the terms »nymphet« for the nine
to fourteen year-old girls to whom the protagonist is attracted and »faunlet« for the young
male counterpart of a nymphet, in the same way that the mythological fauns where the coun-
terparts of the nymphs. A »nympholept« finally is one who could discern nymphets from
other girls. Obviously, the terms apply to real people.
In all these cases, the result is a reorganization: Things that more familiar categories keep
apart are grouped together; things that familiar categories group together are distinguished. If
the reorganization proves useful and reveals important likenesses and differences, the new
categories provided by the literary works advance our understanding of the actual world. Re-
organization leads to new true beliefs and perhaps to new knowledge. But we do not learn
them from the works. They do not teach us that a certain man is a Don Juan and a certain girl
a nymphet; that a certain action is quixotic and that a certain mental process represents a
crystallization process. What the works provide us with are rather the resources for acquiring
Artworks can provide new perspectives on objects that enhance our understanding of them.32
By emphasizing and attenuating, exaggerating and downplaying, adding and omitting, de-
forming and alienating, pictures make us aware of hitherto unnoticed features of objects
thereby yielding a new way of conceiving of them. Edward Weston’s »Pepper«-photographs
(fig. 1) highlight hitherto overlooked likenesses and differences between the form and texture
of paprika peppers and human bodies hence suggesting a new way of looking at both of them.
Claude Monet’s Gare St-Lazare (fig. 2) focuses on the visual impact of sunlight, smoke and
steam, suppressing the detailed features of the objects we would expect to find in a picture of
a railway station. By inviting us to attend directly to our visual experience rather than seeing
31 Elgin 2002, 3–6; John 2001, 338–339; Scholz 2001, 41–43.
32 See Young 2001, 76.
past it for the sake of the information it provides, the painting suggests a new way of looking
at our surroundings that reveals features of it we usually neglect. Francisco de Goya’s aqua-
tint prints The Disasters of War (fig. 3) focus on the brutality and cruelty of war scenes with
anonymous protagonists rather than known patriots, and do not integrate the isolated scenes
into a narrative that makes sense of them. The prints thereby avoid the bombastic heroics of
most previous war art and provide a new and disturbing perspective on war that is devoid of
the consolation of divine order or the dispensation of human justice. The perspective drives
home the message that there is nothing noble about war, and it may bring us to hypothesize,
for example, that the killing Goya depicted so perspicuously »obeys urges embedded at least
as deeply in the human psyche as any impulse toward pity, fraternity, or mercy«.33
Literary works can function in a similar way. By carefully selecting and describing fic-
tional incidents, actions and characters, they provide perspectives on real people and their
relationships and interactions.34 In Edward Casaubon, George Eliot has grouped together a set
of qualities to define a character with depth and plausibility. We can learn to see a man in
these terms and so discover more about him, even if we are reluctant to metaphorically call
him a Casaubon. David Novitz suggests that convincing resemblances between the clergyman
in Middlemarch and some academics may bring us to entertain a hypothesis about the latter:
»If […] Casaubon resembles some academics in respect of his confined interests and his love
of isolation, and if, as is also the case, Casaubon demonstrates a lack of security that mani-
fests itself in a mean-spirited desire to control others, one might venture the hypothesis that
the confined interests and truncated lives of some academics are indicative of a lack of confi-
dence that, in its turn, is productive of the desire to exercise power and control over other
33 Hughes 2003, 289.
34 See Graham 2005, 70.
35 Novitz 2004, 998.
Fig. 1: Edward Weston, Pepper No. 30, 1930
In all these cases, the works suggest new beliefs or lead to new hypotheses. Weston’s
photographs suggest that paprika peppers, viewed from a certain angle and in a certain light,
are evocative of human bodies. Middlemarch leads to the hypothesis that some academics
demonstrate a lack of security that manifests itself in a desire to control others. The perspec-
tives provided by the works advance our understanding if the beliefs and hypotheses cohere
with established beliefs and enable us to make sense of the phenomenon they concern. If the
hypotheses are thus confirmed, we will eventually believe them. If the beliefs are true, they
may constitute knowledge. But perspectives are not reduced to such knowledge. Weston’s
photographs furnish us with a new way of looking at paprika peppers and human bodies; the
clergyman in Eliot’s novel with a new way of thinking about academics. In both cases, what
we learn resists a settled paraphrase. As in grasping a metaphor, one engages in an open-
ended exploration of salient similarities between two disparate entities, paprika peppers and
human bodies in one case, a fictional character and some academics in the other case. Addi-
tionally, what we learn will, at least in the perceptual case, often outstrip our words.
Artworks can raise important questions that prompt further inquiry. Literary works, for ex-
ample, seldom offer moral doctrines or solutions to moral problems. More often, they show
that a moral decision is more complex and difficult than we thought so far, thereby posing
demanding questions.36 When we engage properly in Gottfried Keller’s Green Henry, for in-
stance, we will find ourselves confronted with moral questions about self-deceit and veracity,
about autonomy and our duties towards others, about the value of individual fulfilment and
the demands of society, and so on. These questions are often as pertinent to our own life as to
the lives of the fictional characters. Raising them does not straightforwardly impart new moral
beliefs, let alone new moral knowledge. The contribution to moral understanding rather lies in
inviting critical reflection and testing moral beliefs we have uncritically adopted at some stage
or other of our lives. The novel achieves this by bringing us to apply our beliefs to complex
imaginary situations that have been described in fine and nuanced details by Keller.
Often, raising pertinent moral questions about some conduct is related to providing a new
perspective on it. By rendering Emma’s adulterous affairs understandable as an attempt to
36 See Novitz 2004, 1001; Scholz 2001, 45.
Fig. 2: Claude Monet, Gare Saint-Lazare,
1877, Musée d’Orsay, Paris Fig. 3: Francisco de Goya, Why?, Plate 32 of
The Disasters of War, 1810–1814
escape the banalities and emptiness of provincial life and withholding any explicit moral
judgment, Madame Bovary provides a new perspective on adultery. The novel thereby raises
questions about adultery and invites the reader to reflect critically on its rights and wrongs.
According to Hans Robert Jauss, this appeal to moral reflection is promoted by stylistic
means. In Flaubert’s style indirect libre, the thoughts of characters are reported without the
signals of direct discourse. The thought concerning adultery expressed in the phrase »at last
she was going to possess those joys of love, that fever of happiness of which she had des-
paired«, for instance, is not explicitly attributed to Emma or to the narrator. Hence, the reader
has to decide whether she should take it as Emma’s or as the narrator’s thought. By employ-
ing this new device and refraining from explicit moral judgment, Flaubert was »able to jolt
the reader of Madame Bovary out of the self-evident character of his moral judgment, and
turned a predecided question of public morals back into an open problem«.37 Not only literary
works raise important questions by providing new perspectives. Elgin suggests that by ex-
pressing a tapestry of mortification, hope, uncertainty, and fear, the Confiteor of Bach’s B-
Minor Mass conveys the utter incomprehensibility of divine forgiveness.38 Given that the sin
to be forgiven is the murder of the forgiver’s son, the music raises the questions: Why should
God forgive? How can God forgive? But according to Elgin, the complex expressiveness of
the music leads to even more basic questions that concern not only divine forgiveness: What
exactly is forgiveness? How is it possible? What does it cost? Again, raising these questions
does not straightforwardly lead to new beliefs or even knowledge. The cognitive contribution
of the music rather lies in problematizing what had previously seemed unproblematic. Real-
izing that we do not really understand forgiveness may prompt further inquiry into issues in
moral psychology that deserve to be examined.
Artworks can provide us with knowledge of how it is (or was or would be) like to have certain
experiences or emotions, or to be in a certain situation.39 They do so by broadening our expe-
rience in encompassing things we might never otherwise have undergone or felt. From Claude
Simon’s novel The Flanders Road we can learn something about how it is (or was) like to
take part in the French military collapse as a cavalry man in 1940. The novel achieves this by
focusing on the perspective of the protagonist and eschewing all explanations from an omnis-
cient narrator. The seemingly chaotic arrangement of the text and the very long and intricate
sentences with their unconventional punctuation evoke the turbulent flood of sensations,
emotions and memories as they show up in the consciousness of the protagonist: jumbled,
discontinuous, fragmented and in apparently random juxtaposition. What we learn by reading
the novel can constitute phenomenal knowledge and deepen our understanding of some
aspects of Second World War if it fits with our experience and well-established beliefs. The
acquaintance that leads to this knowledge is, of course, symbolically conveyed and imagina-
tive rather than a direct acquaintance through experience or feeling. We fortunately neither
directly experience the bombardment by German airplanes that killed almost the whole cav-
alry regiment, nor do we feel the horror and fear of the protagonist. Rather, Simon’s detailed
description combined with our own remembered experiences and feelings of horror and fear
enable us to sympathize with the protagonist and imagine what it is like to be so heavily bom-
37 Jauss 1986, 180–182.
38 Elgin 2002, 10.
39 See John 2001, 333–335; Novitz 1987, 132–137.
barded. But although the phenomenal knowledge we may gain thereby is based on verbal de-
scriptions, it is not reduced to propositional knowledge. The descriptions enable us to imagi-
natively acquaint with the wartime experiences of a cavalry man; but they do not capture the
whole content of our acquaintance.
The cognitive function of providing phenomenal knowledge is not confined to literary
works. Paintings and music, for instance, can achieve the same by non-verbal means. From
Rembrandt’s Bathsheba with King David’s Letter (fig. 4) we can get an idea of what it is (or
would be) like for a woman who loves her husband to be forced into adultery by an attractive
man of power. The painting achieves this by, among other things, showing the expression on
Bathsheba’s face that, according to Berys Gaut, reveals »a mixture of introspective sadness,
tinged with a gentle resignation, with perhaps a hint of a half-smile of erotic anticipation«.40
From the second movement of Beethoven’s Eroica, and from the »Dead March« from Han-
del’s Saul we can learn what it is like to experience various shades of mourning. The two
pieces of music achieve this by providing us with different perspectives on mourning. The
»Funeral March« from the Eroica with its large, unrestrained gestures is unsettling; the
mourning it expresses is tied up with distracted and passionate railing against fate. The »Dead
March« from Saul is stately and dignified; the mourning it expresses is restrained, dignified
and resigned.41 In all these cases, the works may lead to propositional knowledge. But the
phenomenal knowledge gained by the works does not reduce to it. For instance, if we thought
that mourning is always passionate, bitter and tinged with despair, we can learn from the
»Dead March« that mourning can be restrained, dignified and resigned. But this propositional
characterization does not fully capture what it is like to experience the mourning expressed by
Art can contrive thought experiments. It has been suggested that while scientific and philo-
sophical thought experiments are fictions in science and philosophy, literary fictions are
40 Gaut 2007, 22.
41 Young 2001, 93–96.
Fig. 4: Rembrandt, Bathsheba with King
, 1654, Musée du Louvre, Paris
thought experiments in art.42 Hence, if thought experiments make cognitive contributions,
works of fiction do as well. Like scientific and philosophical thought experiments, literary
fictions ask what would happen if we assume that certain conditions obtained and invite us to
explore the consequences of making these assumptions. They advance our understanding if
their driving assumptions are adequate. Some present a paradigmatic case, others go to the
extremes. In both cases, they may afford epistemic access to things that are obscured in the
less pure instances we typically encounter in reality.
Like ordinary thought experiments, literary fictions can be used to support, undermine,
illustrate or flesh out some hypothesis or idea. William Styron’s Sophie’s Choice brings us to
imagine a situation in which the Jewish heroine was given a choice by a drunken medical
doctor on her arrival at Auschwitz: she had to choose which one of her children was to be sent
immediately to the gas chamber, or they would both be sent there. By rendering the details of
her choice extraordinarily vivid and exploring its consequences for Sophie’s life, the novel
supports the hypotheses that there are genuine moral dilemmas and that the guilt someone
feels from making such a choice can obsess and even destroy her.43 George Orwell’s 1984
invites us to imagine a society where the mighty party regularly revises historical records to
concord the past to the current political agenda and destroys all previous records. Besides dis-
crediting and eliminating bits of evidence, the party also sets arbitrary standards for what
counts as evidence and reason by convincing people that they can and should embrace contra-
dictions. The novel can thus be read as an epistemological thought experiment that »under-
mines the conviction that intersubjective agreement, grounded in what is intersubjectively
agreed to be good evidence, suffices for epistemic justification«.44 Jane Austen’s Pride and
Prejudice imagines a landed gentry society of early 19th century England, seen from the
viewpoint of a young woman with a tendency of judging the character of other persons on
first impressions. By showing how she is forced to revise her judgments as the story proceeds,
the novel illustrates the hypothesis that first impressions are a poor guide to a person’s char-
acter. Heinrich von Kleist’s Michael Kohlhaas incites us to imagine a man with a distinctive
sense of justice who has been wronged. Since he remains unsuccessful in demanding his
rights through legal channels, he resorts to criminal means and initiates a private war. The
novella fleshes out the idea that one is entitled to take the law in one’s own hands if one’s
rights are not satisfied by the state, and it shows where the acceptance of such an idea would
There are, of course, differences between ordinary thought experiments and literary fic-
tions. Firstly, while scientific and philosophical thought experiments are usually quite austere,
literary fictions are developed in much more detail. Furthermore, many different aspects of a
literary fiction matter, and every semantic distinction can make a difference to what the fic-
tion conveys. Not only the plot, but the characters, their perspective on events, the perspective
of the narrator, even the sound, texture, tone and grain of the descriptions may be significant;
and the subtlest details of what the fiction entices us to imagine may be relevant for what it
transmits.45 As a result, literary thought experiments are often more convincing than their
philosophical counterparts; but to get their lesson, considerably more interpretative effort is
42 Elgin 1996, 180–183; 2002, 8–12; 2005, 47–53; cf. Carroll 2002, 3–26; John, 1998, 332; Swirski 2007, ch. 4.
Carroll seems wrongly to assume that the knowledge thought experiments lead to is always conceptual
43 See Gaut 2007, 162–164.
44 Elgin 2005, 50–51.
45 See Elgin 2005, 48–49.
needed. Secondly, in contrast to scientific and philosophical thought experiments, literary
fictions are usually not contrived as part of a theory and not even with reference to a specific
theory or hypothesis.46 As a result, the hypotheses are usually not explicitly stated in the work
but have to be inferred from it. In most cases, there will be a variety of hypotheses that can be
correlated with a work; and new interpretations can always reveal further ones. Sophie’s
Choice, for instance, is about many other things than moral dilemmas, for example, about the
Nazi death camps, about the nature of absolute evil, about what it means to lose faith in God
and about the love of a young, self-congratulatory man to an older, much more experienced
woman. Concerning all these further topics the reader can infer hypotheses the novel sup-
ports, undermines, illustrates or fleshes out.
The two differences between ordinary thought experiments and literary fictions are related
to the fact that while scientific and philosophical works may contain and use thought experi-
ments, literary fictions are thought experiments. Hence, it becomes evident that contriving
thought experiments is not on the same level as are the other cognitive contributions of art-
works discussed in this Section. Rather, by functioning as a thought experiment, a fictional
work can provide a new perspective on a topic; and by providing a new perspective a work
can raise pertinent questions about the topic, deliver new categories for dealing with them and
teach us what it is like to be in a situation relevant to the topic.
2.2 Contributions that Reach Beyond Knowledge
Besides cognitive contributions that do not classify as knowledge, artworks make contribu-
tions to our understanding that go beyond a knowledge of isolated facts.
Artworks can deepen our understanding by enabling us to grasp connections between what we
already believe. Let me illustrate this by a slightly modified example I owe to Noël Carroll.47
Lorraine Hansberry’s play A Raisin in the Sun portrays a few weeks in the life of the Young-
ers, an African-American family living on the South Side of Chicago in the 1950s. When the
play opens, the family is about to receive an insurance check over $ 10,000. Each of the fam-
ily members has a different idea on what to do with this money. As the play progresses, they
clash over their competing dreams. At a certain point, Mama places a down payment on a
house for the family. This house is in an entirely white neighbourhood. When the future
neighbours find out that the Youngers are moving in, they offer them money for staying
Suppose that the white audience in 1959 already believed the following two propositions:
(3) Persons deserve equal treatment.
(4) African-Americans are persons.
Assume furthermore that these beliefs are pale in the sense that the audience does not fully
appreciate their import. Now, the play makes (4) clear by showing that the dreams and family
bonds of the major black characters are no different from those of other persons. It thereby
46 There are exceptions. According to Carroll, there can be little doubt that Ursula Le Guin’s short story »The
Ones Who Walked Away from the Omelas« is designed to challenge utilitarianism (Carroll 2002, 21, note
34); the same may be true of Charles Dickens’ Hard Times (cf. Young 2001, 85, 100). But in both cases, this
is hardly the only aim the authors pursue with their fictions.
47 Carroll 1998, 142–143.
prompts the audience to acknowledge the connection between (3) and (4) and to draw the
(5) African-Americans deserve equal treatment.
On this basis, the play encourages the white audience to form the moral judgment that the
way in which the prospective white neighbours of the Youngers respond to their purchasing a
house in their neighbourhood is wrong. But let us focus on (3), (4) and (5). In one sense, the
audience acquires the new belief that (5) holds. But since (3) and (4) imply (5), the audience
is already committed to (5). In another sense then, the audience deepens its understanding of
what it already believes.
Improving Cognitive Abilities
Artworks can enhance or refine our general cognitive abilities of reasoning, emotion, percep-
tion, imagination, memory, and so on.48 They do so by providing us with exercises in cogni-
tive activities, or by representing exemplars thereof. In one case, we learn by doing. A novel
can improve our cognitive abilities by prompting us to engage them reflectively in our attempt
to come to grips with the development of the characters and their interactions. In the other
case, we learn by encountering an exemplar. A novel can advance our cognitive abilities by
carefully describing how the characters of the story make use of these abilities and refine
them during their development. In both cases, the improvement of a general cognitive ability
is achieved by improving a specification of it with respect to a certain problem or phenome-
non. A novel may improve our moral reasoning by refining our ability to morally reason
about the problem of illegal immigrants; a series of paintings may improve our perception by
refining our ability to perceive the impact of different lightening conditions on the atmosphere
in a certain situation. Having one’s cognitive abilities improved is not identical to having
one’s stock of knowledge increased. The improvement of cognitive abilities may, of course,
lead to new knowledge. But it may equally enable us to construe adequate category schemes,
develop useful diagrams, maps and models, raise insightful questions and contrive helpful
idealizations and thought experiments.
Since the improvement of general cognitive abilities by engaging with artworks is widely
discussed,49 I focus on the enhancement of the more specific ability to apply certain principles
or concepts. This cognitive contribution is important since we often possess general principles
or concepts that are very abstract and that we may not be able to apply to particular situations.
As we have seen, having such an ability is part of our understanding. Hence, its enhancement
advances our understanding; but it does not necessarily lead to new knowledge. Artworks can
improve our ability to apply general principles and concepts precisely because they deal with
the concrete and particular rather than the abstract and general.50 Consider two cases. In non-
challenging cases, artworks supply us with vivid – often imaginary – examples. They thereby
enable us to see how to apply the abstractions to particulars, making the principles or concepts
more vivid; often not only by revealing their import, but also by providing the correspondent
48 Besides cognitive abilities or skills that have to do with processing information about the world, artworks
may also enhance or even provide strategic abilities or skills. A favourite hero may furnish us with purely
practical strategies for handling a tricky situation; strategies which we may adopt when we find ourselves in a
similar situation (Novitz 1987, 119; 2004, 995).
49 Hilary Putnam (1978) focuses on improving practical reasoning, Jenefer Robinson (1997) on educating our
emotions, Berys Gaut (2006) on enhancing our imaginative capacities, Martha Nussbaum (1990) discusses
the refinement of a whole range of cognitive and moral abilities.
50 See Carroll 1998, 144–146; Conolly/Haydar, 2001, 115; Young 2001, 95.
knowledge of how it is like to be in such a situation. Pride and Prejudice, for instance, gives
us an arresting example by which to understand the general proposition that first impressions
are a poor guide to a person’s character. Goya’s Disasters of War (fig. 3) vividly exemplify
the horrors of war. In challenging cases, artworks supply us with complex imaginary situa-
tions or hard cases. They thereby drive us to test and, if necessary, revise our principles or
concepts, thereby refining our understanding. Flaubert’s Madam Bovary questions the as-
sumption that adultery is morally wrong by describing Emma’s adulterous affairs in all their
different aspects and complexities. Orwell’s 1984 undermines the conviction that intersubjec-
tive agreement suffices for epistemic justification by describing a plausible case where such
agreement is generated by epistemically illicit means. Of course, the same artwork can present
non-challenging cases with respect to certain aspects of its content and challenging cases with
respect to other aspects. While the challenging cases may lead us to new beliefs and even to
new knowledge, the non-challenging cases do not. But they nevertheless deepen our under-
standing of our principles or concepts by teaching us how to apply them to particular situa-
tions and making them more vivid.
3. Non-Cognitivist Objections
Here are three of the most pressing non-cognitivist objections against the epistemic claim of
(6) No-belief objection. Since artworks as such do not refer to the actual world, they as
such do not convey beliefs about it, they at most suggest possibilities. A novel should
be read as starting with »let us imagine that…«, not »it is asserted that…«. To acquire
beliefs about the actual world from artworks requires a refusal of the aesthetic stance,
since this stance involves a suspension of reference to the world.51
(7) Banality objection. If artworks convey beliefs about the world, their content is com-
pletely banal.52 One does not learn these beliefs from the artworks. Having them al-
ready is more of a precondition for understanding the works. The only moral truths
Dostoevsky’s Crime and Punishment conveys are banalities like »murder is wrong«,
which we do not learn from the novel. Comprehending the novel rather presupposes
that the reader grasps the moral precepts that motivate its narrative.53
(8) No-justification objection. Even if one acquires true beliefs from artworks, they can
never be justified or at least not simply on the basis of one’s acquaintance with the
work. Dickens’ Bleak House may be accurate about the slowness of estate litigation in
nineteenth-century Britain, but it cannot provide evidence for its accuracy. One cannot
be justified then in believing its claims simply on the basis of reading the novel. For
that one needs to consult the history books.54 Hence, artworks can at most suggest hy-
potheses or present a point of view.55
All three objections assume, like most non-cognitivist objections, that if artworks have a
cognitive function at all, it is the function of being a source of non-trivial knowledge. They
51 Diffey 1997, 30.
52 Stolnitz 1992, 193–194.
53 See Carroll 1998, 130; 2002, p. 4.
54 Stolnitz 1992, p. 169.
55 See Graham 2005, p. 64.
then argue that artworks as such cannot have this function since according to (6) they violate
the belief condition for knowledge, according to (8) the justification condition for knowledge
and according to (7) they are at best a source of trivialities. As we have seen, the underlying
assumption is false. Artworks can provide new categories and perspectives, raise important
questions, provide knowledge of how it is like to have certain experiences or emotions, de-
velop elaborated thought experiments, enable us to grasp connections between what we al-
ready believe and enhance our cognitive abilities. Hence, even if the three objections would
be successful, they would only undermine some (and, according to many cognitivists, com-
paratively insignificant) cognitive functions of artworks. However, as I will argue in the re-
maining part of this last Section, the objections are not successful.
3.1 No-Belief Objection
The no-belief-objection reflects a shift that is very common in the debate about aesthetic cog-
nitivism: that from art to fiction. But not all artworks are fictional. Non-fictional works refer
to actual objects and convey beliefs about them. Biographies and documentary films with ar-
tistic aspirations refer to real people and places and can be thought of as beginning with »it is
asserted that…«. Many portraits and landscapes denote real persons and scenes and impart
beliefs about them by representing them as having certain properties.
Even fictional works sometimes refer to actual objects and convey beliefs about them.
Some novels consist partly of fictional and partly of non-fictional statements. T.C. Boyle’s
Water Music contains long descriptions of the environmental conditions in West Africa. Other
novels include non-fictional elements within fictional descriptions. Keller’s Green Henry
opens with a description of shipping on the Lake of Zurich. Its sentences are simultaneously
fictional statements »about« a fictional world and factual statements about a real town and
lake. They thereby convey beliefs about how Zurich and its lake looked in the nineteenth
century. In a comparable way, Konrad Witz’ representation of St. Peter’s encounter with
Christ after the resurrection in The Miraculous Draught of Fishes (fig. 5) locates the fictional
subject at the Lake of Geneva with the Mont Salève in the background. The painting thereby
imparts beliefs about how the lakeside looked in the fifteenth century (and suggested to Witz’
contemporary compatriots that the miracle occurs here and now).
Moreover, like thought experiments in science and philosophy, fictional works can con-
vey beliefs about actual people or situations by highlighting features they share with them but
are not easily accessible in them. The claims that lead to the beliefs in question are implicit in
Fig. 5: Konrad Witz, The Miraculous Draught of
Fishes, 1444, Musée d’Art et d’Histoire
the works rather than explicit claims the works put forward. In this way, Orwell’s 1984 claims
that intersubjective agreement is not sufficient for a belief to be epistemically justified; Aus-
ten’s Pride and Prejudice claims that first impressions are a poor guide to character. The no-
belief objection, thus, presupposes an over-simple distinction between claims about the actual
world and claims about mere possibilities or counterfactual states of affairs.56 As thought ex-
periments show, explorations of counterfactual states of affairs can reveal much about the
actual world. And claims about some features of the world, such as virtues, values and duties,
entail or otherwise ground claims about counterfactual conditions, whose features can be ex-
plored through thought experiments. A claim about having a certain virtue, for instance,
commits me to a claim about how I would behave in counterfactual situations. If I hold that I
am truly courageous, I must also claim that I would withstand certain sorts of hardship I have
never been subject to. So, it is relevant to determine whether I am courageous to figure out
what I would do in these merely possible situations. Hence, we can learn about aspects of the
world through imagination, and ordinary or literary thought experiments can aid these imag-
Finally, the aesthetic stance does not involve a suspension of reference to actual objects.
This is shown by our aesthetic evaluations of artworks. The evaluation of the accuracy of de-
scriptions and the faithfulness of representations is often important to an aesthetic evaluation
of artworks.57 This is true of non-fictional works, such as biographies, documentary films,
portraits and landscapes, where serious falsities and misrepresentations of their subject matter
is an aesthetic flaw. To say that a biography is unfaithful to the facts or that a documentary
film distorts the truth is clearly a legitimate aesthetic criticism. But also in the case of fictional
works, the evaluation of their explicit or implicit claims may be aesthetically relevant. Or-
well’s 1984, for instance, would certainly have been of less aesthetic value if it would have
claimed that being acquired between two and three in the afternoon is sufficient for a belief to
be epistemically justified.
3.2 Banality Objection
The second objection claims that the beliefs artworks convey are banal; and that we do not
learn them from the works since having them already is a precondition of understanding the
works. But artworks can provide us with interesting general beliefs. The mentioned belief we
can get from Orwell’s 1984 is certainly not banal. From Styron’s Sophie’s Choice we can ac-
quire the beliefs that there are genuine moral dilemmas and that the guilt someone feels from
making a choice in a dilemma situation can obsess and even destroy her. Whatever the truth
of these beliefs may be, they are not banal. The existence of moral dilemmas has been denied
by many philosophers; and they have attempted to explain away feelings of guilt as irrational
The general claims an artwork makes are typically made implicitly by the work’s treat-
ment of particulars. They are displayed in the fine-grained descriptions or representation of
particular characters and events. Hence, the cognitive contribution of artworks does usually
not primarily consist in conveying general beliefs considered in abstraction from the particu-
lars of the narrative and its characters. It rather lies in the detailed descriptions or representa-
tions of particular (and often imaginary) cases that suggest the general beliefs. The cognitive
56 See Gaut 2006, 117; 2007, 147–148.
57 See Gaut 2005, 447.
58 See Gaut 2007, 162–164.
contribution of Austen’s Pride and Prejudice does not primarily consist of the claim that first
impressions are a poor guide to character; it rather lies in the nuanced description of an im-
aginary case that, among other things, illustrates this claim. The cognitive contribution of
Goya’s Disasters of War (fig. 3) does not primarily consist of the message that there is noth-
ing noble about war; it rather lies in the unadorned perspective on war they provide, and that,
among other things, suggests this message.
Furthermore, that having certain beliefs is a precondition of comprehending an artwork
does not show that we cannot acquire new interesting beliefs from the work. Comprehending
a scientific treatise, for instance, presupposes a vast number of scientific, semantic and every-
day beliefs; but, of course, we can nonetheless acquire interesting beliefs from reading the
treatise. The same holds for artworks. We indeed do not learn that murder is wrong form
Crime and Punishment; and having this belief may be a precondition of actually compre-
hending Dostoevsky’s novel. But this does not show that we cannot learn other, typically
more specific, truths from the novel, for instance, about what it is like to live with having
committed murder.59 Moreover, an artwork can even undermine some of the beliefs its com-
prehension presupposes. Understanding Oedipus at Colonus may presuppose the conviction
that every person who has committed parricide and incest deserves blaming. Sophocles’ play
then undermines this conviction by representing a person who is the victim of forces beyond
his control and has been ignorant of the consequences of his actions; who has suffered terribly
and who feels genuine remorse.60
3.3 No-Justification Objection
There is a strong and a weaker version of the no-justification objection. The strong version
claims that beliefs about the world we acquire from artworks can never be justified. This
claim commits a genetic fallacy. The origin of beliefs in artworks, or even in fictional works,
does not prevent them from being justified precisely the way we justify beliefs derived from
experience or testimony.61
The weaker version claims that we cannot be justified in beliefs we acquire from artworks
simply on the basis of our acquaintance with the works. This claim is partly correct but does
not show what it should. Artworks can provide some justification for the beliefs they convey.
This seems evident in case of non-fictional works; it is, however, also true of fictional works.
Through their choice and presentation of narrative events they often, for instance, suggest
some reasoning the audience is invited to adopt and provide imaginative acquaintance that
grounds beliefs about what it is like to be in a certain situation. But to fully justify the beliefs
acquired from artworks we have to go beyond the works and investigate whether the beliefs
fit with well-established beliefs, helping us to make sense of the phenomena they concern.
This, however, does not prevent us from using the works as sources of knowledge, since
the same holds for undeniable sources of knowledge such as reference works. They can only
be said to impart knowledge about the world if we are justified in believing that they are reli-
able. But our knowledge of their reliability is not acquired from our acquaintance with the
reference work. Hence, there seems no more reason to hold that fictional artworks cannot
yield knowledge than to hold that reference works cannot.62 This may be denied by pointing
out that reference works, but not fictional works, are subject to what Gaut calls an »institu-
59 See Conolly/Haydar 2001, 122.
60 See Young 2001, 85.
61 Novitz 2004, 1002.
62 Novitz 1987, 132; 2004, 999.
tional guarantee«: they are usually refereed, their claims being checked by peer reviewers. So,
if one consults a reference work, one seems to know that it is a reliable source of knowledge;
but that never seems true of fictional works. Now, there is certainly a difference here. But it is
only a difference of degree in the robustness of the testimony. First of all, good art critics may
function somewhat like scientific peer reviewers.63 Furthermore, simply by looking at the
reference work, one can neither tell whether the peer reviewing has really occurred, nor that it
has been done properly. To ascertain whether it has been done properly, one must go beyond
the text and investigate its generative conditions. That again places fictional and reference
works on an equal footing.64
The no-justification objection is formulated in terms of beliefs and hence with respect to
knowledge. But the problem it raises is also pertinent to understanding since it too implies
some sort of justification. Categories, perspectives, questions, acquaintances provided by art-
works and abilities promoted by them can be misleading. Not all categories suggested by
novels prove to be useful; popular films often provide a rather delusive perspective on war; by
many works of fiction we are offered a quite unhelpful sense of what it is like to be in love;
and cognitive abilities can be so misrepresented in artworks that we would be badly misled if
we tried to apply them in real life, even though they apply fruitfully to the world of the work.
But if one can be mistaken, one needs confirmation that one is right. Hence, even the putative
cognitive contributions that do not constitute knowledge or go beyond it need to be justified.
The concept of justification that is required to encompass them is more complex than tradi-
tional truth-conducive justification. There are two reasons why this is so. One is that some of
the putative contributions are non-propositional and hence neither true nor false. The other is
that truth is only one of many epistemic goals that have to be weighed against each other.
Hence, a concept of justification is required that is applicable to non-propositional compo-
nents and related to a plurality of epistemic goals. Elgin has, convincingly to my mind, argued
that the model of wide reflective equilibrium provides a conception of justification that fulfils
both requirements. According to it, a system is justified if it is in wide reflective equilibrium.
Roughly, a system is in wide reflective equilibrium if its propositional and non-propositional
components are reasonably in light of one another and in light of relevant background theo-
ries, the system as a whole is reasonable in light of our antecedent commitments about the
subject at hand and does justice to our epistemic goals.65 The justification artworks provide
makes their contributions initially tenable; to be fully tenable they have to be integrated into a
system in wide reflective equilibrium.
I have argued that an epistemology of understanding is better suited than a theory of
knowledge to do justice to the cognitive achievements of science, philosophy and the every-
day; and that such an epistemology can and should accommodate a wide range of cognitive
functions of artworks and provides a suitable epistemological framework for aesthetic cogni-
If the epistemic claim of aesthetic cognitivism is established and a suitable epistemologi-
cal framework sketched, reflections on the diverse cognitive functions of artworks can give
63 Young 2001, 106.
64 Gaut 2005, 442–443.
65 Elgin 1996, 101–145.
indications of how to understand certain cognitive contributions or even reveal further contri-
butions that may also play an important role in science, philosophy or everyday life. Such
reflections can thereby advance our understanding of some cognitive achievements or even
change our understanding of what counts as a cognitive achievement. The issues raised by
considering artworks as a source of understanding are then not only important to aesthetics.
They are also of general epistemological interest. This is in part afforded to the fact that what
we learn from artworks often lies in epistemologically challenging domains. Hence, an epis-
temology of understanding that transcends the narrow bounds of traditional theory of
knowledge is well advised to investigate the cognitive functions of art.66
Brown, James R./Fehige, Yiftach 2010, »Thought Experiments«, in: Stanford Encyclopedia of Philos-
ophy, http://plato.stanford.edu/entries/thought-experiment/ [11.7.2010].
Carroll, Noël 1998, »Art, Narrative, and Moral Understanding«, in: J. Levinson (ed.), Aesthetics and
Ethics. Essays at the Intersection, Cambridge, 126–160.
— 2001, »The Wheel of Virtue: Art, Literature, and Moral Knowledge«, in: Journal of Aesthetics and
Art Criticism, 60, 3–26.
Conolly, Oliver/Haydar, Bashshar 2001, »Narrative Art and Moral Knowledge«, in: British Journal of
Aesthetics, 41, 109–124.
Cooper, Neil 1994, »Understanding«, in: Proceedings of the Aristotelian Society, Suppl. 68, 1–26.
— 1995, »The Epistemology of Understanding«, in: Inquiry, 38, 205–215.
Diffey, T. 1997, »What Can We Learn from Art?«, in: St. Davies (ed.), Art and its Messages: Mean-
ing, Morality, and Society, University Park, 26–33.
Ducheyne, Steffen 2006, »The Argument(s) for Universal Gravitation«, in: Foundations of Science,
Elgin, Catherine Z. 1996, Considered Judgment, Princeton.
— 2002, »Art in the Advancement of Understanding«, in: American Philosophical Quarterly, 39, 1–
— 2004, »True Enough«, in: Philosophical Issues, 14, 113–131.
— 2005, »The Laboratory of the Mind«, in: J. Gibson et al. (eds), A Sense of the World. Essays of
Fiction, Narrative, and Knowledge, New York, 43–54.
— 2006, »From Knowledge to Understanding«, in: St. Herrington (ed.), Epistemology Futures,
— 2009a, »Is Understanding Factive?«, in: A. Haddock et al. (eds), Epistemic Value, Oxford, 322–
— 2009b, »Exemplification, Idealization, and Scientific Understanding«, in: M. Suárez (ed.),
Fictions in Science. Philosophical Essays on Modelling and Idealization, New York, 77–90.
Gaut, Berys 2005, »Art and Knowledge«, in: J. Levinson (ed.), The Oxford Handbook of Aesthetics,
— 2006, »Art and Cognition«, in: M. Kieran (ed.): Contemporary Debates in Aesthetics and the
Philosophy of Art, Malden, 115–126.
— Art, Emotion, and Ethics, Oxford 2007.
Glock, Hans-Johann 2008, What is Analytic Philosophy?, Cambridge.
Goodman, Nelson 1978, Ways of Worldmaking, Indianapolis.
Goodman, Nelson/Elgin, Catherine Z. 1988, Reconceptions in Philosophy and Other Arts and Sci-
Graham, Gordon 2005, Philosophy of the Arts. An Introduction to Aesthetics. Third Edition, London.
Hughes, Robert 2003, Goya, London.
Jauss, Hans R. 1986, »Literary History as a Challenge to Literary Theory«, in: H. Adams/L. Searle
(eds), Critical Theory since 1965, Tallahassee.
66 I would like to thank Georg Brun for his helpful comments and suggestions.
John, Eileen 1998, »Reading Fiction and Conceptual Knowledge. Philosophical Thought in Literary
Context«, in: Journal of Aesthetics and Art Criticism, 56, 331–348.
— 2001, »Art and Knowledge«, in: B. Gaut/D. McIver Lopes (eds), The Routledge Companion to
Aesthetics, London, 329–340.
Klein, Peter D. 2005, »Epistemology«, in: E. Craig (ed.), Routledge Encyclopedia of Philosophy, Lon-
don, http://www.rep.routledge.com/article/P059 [11.7.2010].
Kvanvig, Jonathan 2003, The Value of Knowledge and the Pursuit of Understanding, New York.
Moore, G. E. 1903, Principia Ethica, Cambridge.
Novitz, David 1987, Knowledge, Fiction, and Imagination, Philadelphia.
— 2004, »Knowledge and Art«, in: I. Niiniluoto et al. (eds), Handbook of Epistemology, Dordrecht,
Nussbaum, Martha 1990, Love’s Knowledge. Essays on Philosophy and Literature, New York.
Palmieri, Paolo 2005 »›Spuntar lo scoglio piu duro‹. Did Galileo ever Think the Most Beautiful
Thought Experiment in the History of Science?«, in: Studies in the History and Philosophy of Sci-
ence, 36, 223–240.
Plomin, Robert/Craig, Ian 2001, »Genetics, Environment and Cognitive Abilities«, in: The British
Journal of Psychiatry, 178, 41–48.
Pritchard, Duncan 2010, »Knowledge and Understanding«, in: D. Pritchard et al., The Nature and
Value of Knowledge. Three Investigations, Oxford, 1–88.
Putnam, Hilary 1978, »Literature, Science, and Reflection«, in: H. Putnam, Meaning and the Moral
Sciences, London, 83–96.
Riggs, Wayne D. 2003, »Understanding ›Virtue‹ and the Virtue of Understanding«, in: M. DePaul/L.
Zagzebski (eds), Intellectual Virtue, Oxford, 203–226.
Roberts, Robert C./Wood, Jay W. 2007, Intellectual Virtues. An Essay in Regulative Epistemology,
Robinson, Jenefer 1997, »L’education sentimental«, in St. Davies (ed.), Art and its Messages. Mean-
ing, Morality, and Society, University Park, 34–48.
Scholz, Oliver R. 2001, »Kunst, Erkenntnis und Verstehen. Eine Verteidigung einer kognitivistischen
Ästhetik«, in: B. Kleimann/R. Schmücker (eds), Wozu Kunst? Die Frage nach ihrer Funktion,
Stolnitz, Jerome 1992, »On the Cognitive Triviality of Art«, in: British Journal of Aesthetics, 32, 191–
Swirski, Peter 2007, Of Literature and Knowledge. Explorations in Narrative Thought Experiments,
Evolution, and Game Theory, London.
Thomson, Judith J. 1971, »A Defence of Abortion«, in: Philosophy and Public Affairs, 1, 47–66.
Young, James O. 2001, Art and Knowledge, London.
Zagzebski, Linda 2001 »Recovering Understanding«, in: M. Steup (ed.), Knowledge, Truth and Duty,
New York, 235–251.