Start Submission

Reading: Two Originals, One Artwork: On the Ontology of Originals and Improvisations


A- A+
Alt. Display

Research Article

Two Originals, One Artwork: On the Ontology of Originals and Improvisations


Raphael van Riel

University of Duisburg-Essen, DE
X close


There is disagreement as to the ontological status of works associated with an original. Some hold that works like paintings are identical to the concrete particular the artist worked on while creating the artwork. Others suggest that works of this sort cannot be instantiated more than once. In this paper, it is argued that, even if artworks like paintings cannot be instantiated in reproductions, they are nevertheless possibly instantiated more than once. Moreover, it is tentatively suggested that the same holds for improvisations in music. Therefore, even the best candidates on offer do not allow for an identification of the work with a particular or for an identification of the work with a type that is instantiated maximally once. The argument rests on thought experiments, and an inference to the ‘best’ explanation. This inference is, of course, defeasible. Four alternatives, three of which are potentially fruitful, will be identified, and it will be argued that, on balance, these alternatives should be dismissed.

How to Cite: van Riel, Raphael. “Two Originals, One Artwork: On the Ontology of Originals and Improvisations”. Estetika: The European Journal of Aesthetics 59, no. 2 (2022): 119–34. DOI:
  Published on 15 Sep 2022
 Accepted on 11 Aug 2022            Submitted on 29 Sep 2021

I. Introduction

Peter Strawson and others have argued that perfect reproductions of an original should be regarded as instances of the work the original is an instance of.1 Others have resisted this idea and argued the opposite2 or, more radically, have tried to show that at least some originals are identical to the artwork they are associated with.3

We can thus distinguish between two positions regarding the ontological status of works that have an original. Strawson and his supporters deny that originals play an important ontological role. Strawson explicitly argues that we identify originals with the work by virtue of ‘the empirical deficiencies of reproductive techniques’.4 I take this to imply that we would not (and should not) do so if the reproductive techniques were less deficient; the actual deficiencies are contingent (empirical) and, thus, misleading when we consider metaphysical possibilities. So, Strawson seems to embrace at least this weak claim: it is metaphysically possible that some reproduction of an original is an instance of the work the original is an instance of. It is weak in the sense that it does not imply that any actual original could have or could have had reproductions instantiating the same work the original is an instance of.

Nelson Goodman, who assumes that the criteria for an adequate definition of a certain kind of art depend on social practice, regards this practice as grounding the special ontological status that originals have according to this practice. Moreover, he opposes the Strawsonian view, stating that ‘the difficulty of making perfect reproductions of a painting [does not] have anything to do with confinement of the work to the unique original’.5 Rather, it is the fact that the work belongs to the kind of work that allows for forgeries – it is autographic rather than allographic. Peter Lamarque seems to embrace this view given that he takes the original to be identical to the work, because it plays a distinguished aesthetic role.6

These views reflect different considerations about what is relevant for the status of an original. Roughly, one will embrace the claim that originals are essentially the sole instances of the work only if one believes them to have relevant – perhaps aesthetically relevant, though I will not discuss that question here – properties their reproductions necessarily (and not only contingently) lack. Such properties may include perceptually or experientially relevant properties, being produced in a certain era, being an original, or originating from a particular person. If one assumes that reproductions may count as instances of the work, one will do so because one denies that there is an important aesthetic difference between original and (perfect) reproduction, presumably because one assumes that what is aesthetically relevant can be shared by an original and at least one appropriate reproduction.

Note that positions like the one embraced by Goodman somehow lie in the middle. On the one hand, Goodman argues that paintings are instantiated maximally once. On the other hand, he acknowledges that there might be many originals for one work, as in the case of etching. Hence, being instantiated maximally once turns out to be an ‘accidental’ feature of paintings. According to Goodman, it is a feature that does not derive from paintings’ belonging to the category of autographic arts. As Lee Brown puts it, ‘[o]n Goodman’s account, there is nothing inherent about autographic art that requires works of that kind to be either singly or multiply instanced’.7 So it is not the case that pictures are instantiated maximally once because they are autographic. Rather, they are instantiated maximally once because they are produced in the way they are produced.

This paper is an elaboration on this point. In Section II, I argue that, even if a picture cannot be instantiated in reproductions, it is nevertheless possibly instantiated more than once. Accordingly, those who take the status of an original to be relevant need not be committed to the view that works like paintings are either identical to the original or instantiated maximally once. In Section III, I suggest that the same holds for improvisations. The argument rests on a thought experiment, and an inference to the ‘best’ explanation. This inference is, of course, defeasible. Four alternatives to the view defended here will be identified, three of which deserve closer attention. It will be argued that, on balance, these alternatives should be dismissed.

For the sake of simplicity, I will assume that work-identity hinges only on aesthetically relevant properties. In order to exclude reproductions produced by the original author, I take the relation between the object and the author as well as the properties defining the context of production to be aesthetically relevant properties. I thus cover simultaneously two vastly different positions.

Lamarque suggests that

[a]ll works are reproducible, through copies, replicas, photographic reproductions, and in other ways, and the question of work identity rests to a considerable degree on the status and value given to these reproductions. With many paintings seeing even a good reproduction seems (and is) something less than seeing the original of which it is a copy.8

The observation that seeing a good reproduction is less than seeing the original is said to be relevant for the question of work-identity because there is an alleged difference in ‘status and value’ between original and (good) reproduction. On one reading, being an original – and, hence, bearing a particular relation to the author – is aesthetically relevant, though perhaps not itself an aesthetic property.

The opposite view is explicitly endorsed by Eddy Zemach and at least suggested by Jerrold Levinson.9 Here, properties from which aesthetic properties are said to emerge10 include what can intuitively be described as intrinsic or qualitative properties, together with structural properties, such as properties of form and colour.11Artistic properties, in turn, are those not directly perceivable but, rather, contextual (involving, for instance, the position of an artwork within a tradition).12

Both conceptions of aesthetically relevant properties are compatible with the view that works associated with an original are particulars. While different, these positions do not differ on the properties that ultimately determine work-identity. However, they do differ on the aesthetic relevance of these properties and, hence, on the aesthetic relevance of being an original. But questions pertaining to the aesthetic relevance of different types of properties are orthogonal to the questions posed here. So, for the discussion to follow, you may simply pick the criterion you prefer, and either interpret the notion of an aesthetically relevant property literally or as short for ‘aesthetically relevant, artistic, contextual […] property’.

A final note on methodology. The overall goal of this paper is to explore one part – a crucial one – of our aesthetic practice – namely, the conceptual apparatus we employ when considering questions of work-identity and the metaphysical status of artworks. Our practice is more complex. It is partly determined by the actual techniques of production and reproduction, institutions, and customs. In Section II.2.4 I will return to the potential limitations of the argument and related methodological questions later, when the main argument has been spelled out.

II. Two Canvases, One Work

Types of artworks, which actually have one unique original (for example, pictures or sculptures) could be instantiated more than once even if reproductions were not counted as instances of the work.

Intuitively, there could be two originals of the same painting, in some ways similar to the case of printing.13 If this is correct, we can keep the intuition that originals play a privileged aesthetic or ontic role but deny that for a work like a painting to have an original makes it ontologically distinct from artworks belonging to other kinds of art. The argument is designed to persuade (i) philosophers like Goodman who assume that the uniqueness of the original in the case of painting and sculpture is only a ‘contingent’ feature, depending on the specific way the work is created only, and philosophers like Steven Farrelly-Jackson and Lamarque who identify the work with the original. In the following, I will first describe a conceivable situation that, prima facie, leads to the conclusion that there can be two originals of one work (II.1). I will then address a set of possible alternative interpretations of this situation and argue that they seem to fail to adequately describe the example, so that we can shift the burden of proof (II.2).

II.1. The Argument

Consider the following claim: ‘paintings are singly instanced simply because of the physical process by which they are made’.14 The idea is that paintings are instantiated maximally once because they are painted rather than being created by a technique that allows for more than one token. But couldn’t there be a way of producing two originals of the same picture or sculpture? There could.

Assume that Jacques-Louis David were asked to prepare some sketches for a painting of Napoleon. Napoleon is so impressed by one of David’s suggestions that he asks David to paint it twice. Since Napoleon wants originals, and since he knows that our fictional David is capable of painting two pictures simultaneously, one with the left hand, the other with the right, he asks David to produce two maximally similar paintings simultaneously – maximally similar in the sense that they are similar in every logically possible respect. While they will naturally differ with respect to the material used and the specific location in space during the simultaneous production process, they will not differ with respect to perceptual properties, intrinsic physical features, many relational features, and so on. David sets out to produce two instances of one and the same work and succeeds in producing two maximally similar paintings.15

Some comments are in place. Note that we should construe maximal similarity independent of any decision on the issue of work-identity. Otherwise the argument will beg the question. We should thus assume that the paintings are maximally similar in every logically possible respect, and in order to judge whether they instantiate the same work we should consider any relevant truth about the similarity except truths about whether these particulars instantiate the same work. These truths will include a whole range from truths about distribution of colour to truths about the virtuosity of the brushwork.

Given that the conceivability of a logically consistent situation is sufficient for metaphysical possibility (and given that our scenario seems to be innocent in this respect), there remains one open question: is the goal of producing two originals of one and same work accomplished by producing two maximally similar paintings?

My argument hinges upon the assumption that, if two objects do not differ in any aesthetically relevant respect (including author-identity and identity of context of production), then they instantiate the same aesthetic object, for example an artwork. Building on this idea, we can now argue that the two paintings of David’s do not differ aesthetically. This partial condition on work-identity is met on both readings of an aesthetically relevant property mentioned above, given that we have already stipulated that they have the same context of production and the same author. This move depends on giving an extremely broad interpretation of what aesthetic differences might depend on and by showing that the two originals do not differ in these respects.

So, where might possible differences in aesthetic status stem from? Let us focus on works of music to sharpen our intuitions.

To decide whether two performances instantiate the same work of music, the following facts should be more than sufficient:

  • – composer-identity (correct causal link to the same author),16
  • – history of production-identity (the instances must bear a correct contextual or historical link to the process of creation of the work), and
  • – identity of intrinsic features of the instances.

I think that we are safe to apply this interpretation to the case of originals. In the example, the three conditions are met. We have author-identity, the works to which the canvases give access are produced at the same time and in the same context, the process of production has been guided by one and the same intention of the same author, and it has been executed analogously in both cases. Accordingly, no aesthetic difference between the works the canvases give access to could possibly stem from relevant contextual factors. Finally, the works to which the canvases give access are intrinsically type-identical, mirrored by the fact that the canvases are exact physical doubles. Given that these criteria form a sufficient condition for being aesthetically equivalent in the stipulated sense, the originals give access to works that are aesthetically equivalent. Hence, if being aesthetically equivalent is sufficient for work-identity, then the originals are instances of the same artwork – they do not give access to artworks in the plural, but rather to one artwork.

What distinguishes the canvases is merely a set of arbitrary features like position in space or the material used during the process of production. Features like location in space and time seem to be irrelevant in the case of instances of composed musical works, plays, or etchings, and, moreover, it seems implausible to assume that such things matter in any relevant respect for our aesthetic practice or judgements about work-identity.

These considerations entail that David has succeeded in producing two originals of the same work. If this is correct, such works are possibly instantiated more than once, even if reproductions do not count as instances of works which have an original. It now is the opponent’s duty to explain the difference in work-identity based on the seemingly arbitrary differences between the two canvases.

One might feel dissatisfied with this result and hope for a tenable alternative or simply deny that aesthetic equivalence (including author- and context-of-production identity) is sufficient for work-identity. I will thus move to the second part of the argument and consider alternative interpretations of the scenario.

II.2. Alternatives Dismissed

To resist the conclusion that David’s work has two originals, one must accept one of four alternative explanations of our example. I will argue that each of them fails. They are either obviously problematic or the arguments one needs to evoke to defend them are not persuasive. According to the alternative explanations, (i) the paintings are two originals of two distinct artworks, (ii) they are constituent parts of one original, (iii) neither of them is an original, or (iv) just one of them is the original. The last alternative can easily be dismissed. The paintings do not differ in any important respect. The assumption that facts about whether objects of this sort are originals are simply fundamental or, alternatively, indeterminate seems unacceptable.

Interpretations (i), (ii), and (iii) are more promising. Let us consider (iii) first.

II.2.1. Neither Painting Is an Original

An anonymous reviewer brought to my attention that there are occasions on which we would refrain from describing one object as an original simply because another object was created or discovered at the same time by another person. We have such cases of independent discoveries or creations in science and inventions.17 And at least in the case of inventions – the case mentioned by the reviewer – we may speak of originals. If two inventions of the same object, A and B, were to occur simultaneously and independently, we would perhaps refrain from describing any of the two inventions as an original. The case of the two paintings may be similar.

This is an interesting observation, which reveals an ambiguity in our notion of an original. Assume that, contrary to the facts, crossbows had been invented simultaneously in China and in Greece. Which one is the original invention? Perhaps neither. But note that this verdict seems felicitous only when applied to inventions. It does not seem to contradict possible verdicts concerning the corresponding particulars – namely, the individual (first) crossbows in China and Greece, respectively.

The tentative conclusion one may want to draw is that the application of the concept of an original to types, or events of creating or discovering types, is sensitive to things like simultaneous discoveries or inventions and, perhaps, to the entire history. In the case of the crossbow, it may not make sense to say that the Chinese invention of the crossbow is the original one, simply because it has been invented again independently later in Greece. Not so in the case of particulars. The first Chinese crossbow was an original and, if I am not mistaken, nothing that happened simultaneously, later or earlier, bears on this question. After all, neither the first crossbow in China nor the first crossbow in Greece was a copy of some other particular.

This last observation points to the source of the ambiguity. We can use the term ‘original’ in a way that implies that the object so classified is not a copy of something else. When applied to inventions, we indicate that some person or group came up with an idea about a type of object first (or perhaps filed a patent application first). But the question of a copy simply does not arise here. Inventions are things that may have instances. And these instances can be copied. But the type itself cannot be copied. It can be rediscovered, or the action leading to its discovery (or creation) can be repeated.18 Likewise, particulars can be forged,19 whereas types can be appropriated or plagiarized.20 The observation that the following two claims are felicitous seems to be backed by a structural difference:

  1. If the crossbow was invented in China and Greece simultaneously, neither of the two is an original invention.
  2. The truth of (1) does not contradict the idea that there may have been an original Chinese crossbow and an original Greek crossbow.

The structural difference mirrors the difference between copies and forgeries as opposed to original particulars on the one hand, and rediscoveries, appropriations, and plagiarism as opposed to original inventions on the other. But, even if this diagnosis is on the wrong track, the difference in application conditions for the term ‘original’ in the case of particulars as opposed to inventions remain relevant.

These observations tell us something important about the concept of an original. Moreover, they show why the example of two simultaneous inventions does not contradict the idea that, in our example, we have two originals. Both paintings are produced by the artist. Neither extrinsic nor intrinsic features distinguish either one from their counterparts in worlds where just one canvas was produced and which, in this counterfactual world, would count as the original. Compare the case to the particular crossbows: no reinvention or simultaneous invention would bear on the question whether a particular crossbow is an original. Therefore, to argue that neither canvas is an original, one would have to provide some independent reason for accepting (iii) or argue that the observation regarding things like original inventions apply to originals that are particulars as well – despite appearance to the contrary.

I will thus turn to alternative (ii).

II.2.2. Both Paintings Are Constituents of One Complex Original

Why not simply accept that David has produced a painting that consists of two constituent parts? There is some prima facie evidence against this interpretation. First, Napoleon ordered one painting twice. This order seems to be perfectly coherent. If so, and if David succeeded, a coherent description of the result of David’s actions would count each individual painting not as part of one complex painting. Moreover, there appears to be an important difference between someone who saw one third of a triptych and someone who saw just one of the canvases: the former does not know the entire artwork, while the latter does. You may conceive of the situation as follows: one of the paintings is for public display; the other is to be found in Napoleon’s private rooms, to be seen by people close to him only. We could thus assume an established practice, which, I submit, is continuous with our actual practice, which places each painting in a particular context, completely insulated from each other (think, for example, of original, type-identical prints in museums and in private rooms). I thus think the burden of proof rests with those who would want to embrace option (ii). They would be required to explain away the idea that Napoleon’s request is coherent, on a literal understanding. And they would be required to explain why, despite appearances to the contrary, the case resembles, in relevant respects (though obviously not in all respects), the case of a triptych.21

II.2.3. The Paintings Are Associated with Two Distinct Artworks

What about alternative (i) – namely, the idea that the two paintings are or instantiate two artworks? To endorse this alternative, we would need independent reason to assume that works of art associated with paintings are identical to these paintings. However, we cannot simply help ourselves to this assumption without begging the question.

One argument in favour of the claim that the two paintings are associated with two artworks could turn to Nicholas Wolterstorff’s theory of norm-kinds.22 The basic idea is that, if there is no set of criteria fixing what counts as an instance of a work, then the work is a particular. Wolterstorff famously argued that paintings are different from norm-kinds. They are instantiable entities that can have correct as well as incorrect instances. He writes:

[O]bject-works are norm-kinds, and being such they have associated with them certain requirements for something’s being a correct example of the work. What is different in the case of paintings is that there are no such associated requirements. There simply are no requirements for something’s being a correct example of some kind of which The Odalisque is the premier example.23

Wolterstorff goes on to argue that, in the case of ordinary paintings, there are no criteria by virtue of which a canvas can count as a correct example of a work.24 This idea seems to support the view that ordinary paintings differ from, say, works of music in this specific respect. But does this suffice to show that the two originals are not instances of the same work? Whether or not an artwork is a norm-kind (of which the possibility of incorrect instances is a distinctive feature) is logically independent of the claim that paintings can have more than one original. Artworks associated with concrete paintings might be types that do not allow for incorrect instances. Thus, Wolterstorff’s thesis is logically compatible with the claim that David has succeeded in producing two originals of one and the same artwork.

Shifting the burden of proof, we can leave it to the proponent of alternative (i) to provide an argument for why these two originals should be regarded as originals of different artworks. This could be done in either of two ways. She could give general and independent reason for why paintings must be particulars or be instantiated maximally once. Alternatively, if she buys into the assumption that aesthetically (and perhaps additionally author- and context-) equivalent objects instantiate the same work, she could show that the seemingly arbitrary differences between the paintings such as spatial or temporal relations are relevant for work-identity. If our opponent assumes that work-identity hinges on aesthetic properties only, she would have to show why these properties are aesthetically relevant.

II.2.4. The Limits of Our Practice

Let us stay, for a moment, with Wolterstorff, who happily admits that The Odalisque, like any other painting, is an instance of several different kinds or types.25 Now, granting that, we commit to the view that there is a kind or a type that is individuated by all and only those aesthetically relevant properties that both originals share (including or in addition to properties concerning context of production and author-identity). This kind or type, I submit, should be regarded as the work. It is similar to norm-kinds in that its relevant properties are exactly those by virtue of which the instances count as relevantly indistinguishable objects which instantiate the work. It may be different from a norm-kind in that it does not allow for incorrect instances. But this raises the following worry: are the standards for counting objects as instances of a work not the result of an established practice, which, in some sense, is clearly lacking when considering scenarios like the David case? At this point, we should return to the topic of methodology.

The goal of this paper is to study one part of our practice only – namely, our conceptual scheme insofar as it is concerned with questions of work-identity and the ontology of artworks. There is naturally no established practice with respect to David-like cases. And we may conceive of rather different possible, though non-actual practices for each of the alternatives (i)–(iv). Some of these practices would be more rational than others. A practice according to which both paintings are seen as constituent parts of a more complex artwork understood as the original is conceivable, and people engaging in this practice should not be blamed for behaving irrationally. In contrast, at least some practices that would yield a version of alternative (iv) seem rather absurd, such as the practice of privileging the painting that was painted with the left hand, or, say, the painting first seen by a critic, or, in a case of perfect symmetry, assuming that there is a fundamental fact of the matter as to which is the original without assuming any differences between them, given that this cannot be known. But this paper is not a paper about possible practices; it is a paper about our actual practice. And, although we do not have an established practice that is concerned with David cases, for the simple reason that there are none, we do have an established practice with respect to originals and painters, which, I assume, commits us to a certain stance on possible David cases. Judging from our practice, we should accept that David managed to create two originals of the same work. Unlike alternatives (i)–(iv), this view is in consonance with our established practice. We do not accept that artwork status is ungrounded or that it hinges on arbitrary features like location in space, when we keep fixed context and time of production as well as authorship. Hence, we can dismiss alternative (iv). The status of particulars as an original does not hinge on what happens simultaneously to other particulars; hence, we can, based on considerations about our practice, dismiss alternative (iii). We do have an established practice concerning originals that have constituent parts, even canonical rules to be followed in some cases, like that of a triptych. But this practice does not seem to fit the scenario of the two canvases produced by David. Hence, our practice does not yield the result that David produced one complex original that both canvases are constituent parts of. Moreover, part of our practice is determined by what we can coherently say about works, and the corresponding intentions we form. Napoleon’s request seems felicitous. We can, for the time being, dismiss alternative (ii). And, finally, our practice does not commit us to the view that David produced two originals of two artworks; considerations about norm-kinds do not indicate that he did. Hence, we can safely dismiss alternative (i).

Accordingly, one might be tempted to draw the conclusion that every artwork that possibly has an original can have (or could have had) many originals. However, this move is problematic. The example I have given does not entail anything about artists’ having the initial intention to produce maximally one original. If such intentions play a crucial role for work-individuation, then it seems impossible for an artist who has the intention to produce only one original to produce two originals of the same work. Thus, we should describe the situation as follows: it has been shown that having an original, even if exemplified by artworks like paintings, does not imply that the corresponding work is a particular (or instantiated maximally once). Accordingly, if we accept that originals have a privileged status, it is not due to the deficiencies of the actual techniques of reproduction, as Strawson put it,26 but rather due to the deficiencies of our artists that we always have just one instance in the case of paintings and non-cast sculpture.

One may be dissatisfied, because the argument rests on an examination of only one part of our aesthetic practice. One may also believe that I have overestimated the importance of purely conceptual issues. Yet it is still possible to concede that the conceptual apparatus I have employed plays at least a contributing role when it comes to questions of work-identity and the metaphysics of artworks. To the extent that it does, the results of the previous discussion should be taken into account in more generalist approaches to our aesthetic practice.

We are now in a position to apply the structure of the above example to improvisations.

III. Two Sound Events, One Improvisation

The most elaborate discussion of the aesthetic status of improvisations can be found in a paper by Brown.27 His aim is twofold: to offer a criticism of Goodman’s distinction between allographic and autographic arts based on considerations about improvisations and to develop an account of the aesthetic evaluation of improvised works. According to him, what is required to respond aesthetically to an improvisation in an informed way gives us a hint towards the aesthetically relevant properties of improvisations. From this position, Brown concludes, roughly, that improvisations transcend the distinction between allographic and autographic arts because they are neither forgeable in an interesting sense (and thus are not autographic) nor repeatable (and hence conflict with some of the conditions Goodman proposes for allographic arts). The basic insight the paper offers is that an informed response to an improvisation requires the listener to be aware of the fact that what is perceived is created in the course of the production of the aesthetically relevant sound event.28 I will not take issue with the theses defended in the paper, partly because they form a topic not relevant for the present paper and partly because I wholeheartedly agree with most of what is said. However, there is a third conclusion Brown draws in passing that concerns the ontological status of improvisations, nicely summarized in the following passage:

An improvisational action is an aesthetic singularity. If [two maximally similar improvisation events produced by two musicians, H1 and H2] really are improvisational in character, then each harbors its own generative act. Essential to H1 is its being this spontaneous action; essential to H2 is its being that one. […] While H1 and H2 do token many common action-types, e.g., playing middle C, there is simply no type – improvising this – of which they could be tokens. […] With improvisational music, the performance process gnaws into the very essence of the aesthetic object, so to say.29

Brown draws the conclusion that improvisations cannot be reidentified.30 This is not due to our epistemological deficiencies but rather to the fact that improvisations are spatio-temporal processes. However, as we shall see, an improvisation can occur twice, even if it may be impossible to repeat it at a time later than the time it initially occurred (a topic not discussed in the present paper).

I shall take for granted that a musical improvisation is identical to or is instantiated in a sound event, which may be partly individuated by being the outcome of an intentional, bodily process. And it is relatively easy to conceive of a possible (though utterly artificial) situation in which a person produces simultaneously two sound events, which are or instantiate improvisations. Moreover, the interpretations according to which these events do not count as instances of the same improvisation fail for reasons similar to the reasons given in the case of the two canvases produced by our fictional David. Accordingly, we can conclude that improvisations are instantiable more than once.

Imagine a future society that institutes general equality among men and women. At some point, a law is passed according to which anyone who violates general equality is punished brutally. From the law it follows that whoever presents art (be it music or painting, theatre or literature) has to expose it to members of the society regardless of gender. However, some timeworn traditions – say, of some influential sub-group in this society – lead to another law, which is passed at the same time. According to this law, women and men are not allowed to share aesthetic experiences, except for those that are required for (or are at least side effects of actions undertaken for the sake of) reproduction, in the following sense: it is immoral (and, in fact, punished as brutally as violations of laws concerning general equality) for men and women to attend to the same instances (sound events or paintings, sculptures or performances) of an aesthetic object at the same time. Until then, aesthetic practice has been identical to the actual practice associated with galleries, museums, concert halls, and jazz clubs. Now, museums and galleries open from Tuesday until Sunday, three days being reserved for men and three days being reserved for women, and so do concert halls. Jazz clubs, however, face a problem. Improvisations, we assume, cannot be repeated at different times. Imagine the situation jazz club owners find themselves in. If you allow men or women only, you will be punished because you violate laws of general equality – either men or women would be unable to attend the specific improvisation improvised that evening. If you open for men and women, you will be punished equally brutally, because you allow men and women to share an aesthetic experience in the relevant sense. In order to comply with the law, you would have to do both at once: give access to both men and women to the same artwork, while ensuring that they do not appreciate the same instance of an artwork at the same time. And this seems simply impossible.

Now, the owner of a jazz club has a brilliant idea: she builds a sound insulating wall, partitioning her club into two rooms, one for men, the other for women. On the stage, there is one hole in the wall. The first evening the club opens, a musician (a drummer, say – it will work for musicians only who are able to play two type-identical instruments simultaneously) is placed in the whole, left half of the body on the left-, right half of the body on the right-hand side of the wall. On each side of the wall, there is a drum. The musician can now improvise a one-hand-piece – a piece of a type she would have played under different circumstances in a normal room using one hand only – on both sides of the wall. She is able to produce and to improvise two intrinsically identical but numerically distinct sound events. It seems as though, contrary to what Brown claims, there is some ‘type-improvising […] of which [both sound events] could be tokens’. This observation does not – again, contrary to what Brown suggests – contradict his correct observation that ‘[w]ith improvisational music, the performance process gnaws into the very essence of the aesthetic object, so to say’.31

This may, of course, raise the question of what individuates the performance process. I submit that we should construe the case as a case where the musician has one complex action plan and one prior intention to perform a one-handed improvisation twice, involving her ability to produce two equivalent sound events, using each of her hands for one of the two distinct sound events. Each event counts as a performance of the improvisation.32

Now, the legal background scenario becomes relevant. In order to avoid prosecution, the owner of the club defends herself as follows: the audiences do attend to two distinct sound events and hence the law reflecting some moral code has been observed. Yet they also attend to the same artwork, the same aesthetic object. Hence, there is no conflict with the law of equality.

The jazz club owner could thus present an argument that resembles the argument proposed above. She may reason as follows. Assume that the two laws that create trouble had not been passed. How would one judge the situation? We have got two events, each of which gives access to works with the same aesthetic features, produced by the same artist. None of the sound events is privileged. Although these events form parts of a more complex event, the sub-events are the bearers of those properties relevant for a one-hand improvisation. Both men and women attend to the artwork. Moreover, the two events are aesthetically indistinguishable, even if we take context, authorship, and intrinsic features into account. The case seems to be relevantly analogous to the case of David’s originals. Thus, in principle, an artwork that is associated with an improvisation can be instantiated more than once. As a consequence, there is no conflict with the law.

One may worry that I have, by providing a particular legal context, changed the relevant aesthetic practice. But it should be noted that the scenario is construed in a way where the actual practice is still in place and exercised until the laws are passed. It is against this background that the jazz club owner defends her strategy. So, the result does not concern a different practice but, rather, our actual practice. The background has been chosen to make the question at stake more pressing (the same type of performance(s) under normal conditions might inspire an interpretation according to which the whole setting is some kind of performance art; not so in the scenario).

Thus, in this weak sense of being reidentifiable, according to which reidentification is not tied to being reidentifiable at different times, improvisations are reidentifiable, at least in principle. Accordingly, the fact that more complex improvisations, involving other instruments, can be instantiated just once is due to the ontology of the agent; the category of an improvisation is irrelevant in this respect.

IV. Conclusion

We have seen that neither the concept of an original painting nor the concept of an improvisation precludes the possibility of being multiply instantiated. If conceivability, in this specific case, implies metaphysical possibility, then it is metaphysically possible that improvisations and paintings are instantiated more than once. At least in this sense they are ‘repeatable’ or ‘reidentifiable’. This conclusion is preliminary in the sense that I have not ruled out all alternative interpretations. But, for now, the burden of proof rests with the opponent.


1Maria E. Reicher, Zur Metaphysik der Kunst: Eine logisch-ontologische Untersuchung des Werkbegriffs (Stuttgart: dbv, 1998), 30; Eddy Zemach, ‘No Identification without Evaluation’, British Journal of Aesthetics 26 (1986): 239–351; ‘How Paintings Are’, British Journal of Aesthetics 29 (1989): 65–71; ‘Art and Identity’, British Journal of Aesthetics 31 (1991): 363–68; Günther Patzig, ‘Über den ontologischen Status von Kunstwerken’, in Identität und Existenz, ed. Reinold Schmücker (Paderborn: mentis, 2003), 114–15; Anthony Ralls, ‘The Uniqueness and Reproducibility of a Work of Art: A Critique of Goodman’s Theory’, Philosophical Quarterly 22 (1972): 1–18; Peter Strawson, Individuals: An Essay in Descriptive Metaphysics (London: Methuen, 1959), 231. 

2Robert Hopkins, ‘Aesthetics, Experience, and Discrimination’, Journal of Aesthetics and Art Criticism 63 (2005): 119–33; Nelson Goodman, Languages of Art: An Approach to a Theory of Symbols (Oxford: Oxford University Press, 1969). In the case of Goodman, things are less clear. As Hopkins conclusively argues, Goodman should have accepted that perfect reproductions do not differ aesthetically from the originals they are reproductions of. Roughly, Hopkins suggests that Goodman allows for two objects to differ aesthetically only if, at least in principle, one could come to learn to distinguish these objects (based on perceptual information alone). For perfect reproductions, this is impossible. 

3Peter Lamarque, ‘Work and Object’, Proceedings of the Aristotelian Society 102 (2002): 141–62; Steven Farrelly-Jackson, ‘Fetishism and the Identity of Art’, British Journal of Aesthetics 37 (1992): 138–54; Nicholas Wolterstorff, ‘Toward an Ontology of Art Works’, Noûs 9 (1975): 115–42; Works and Worlds of Art (Oxford: Oxford University Press, 1980). 

5Goodman, Languages of Art, 196. See, however, note 2. 

7Lee B. Brown, ‘Musical Works, Improvisation, and the Principle of Continuity’, Journal of Aesthetics and Art Criticism 54 (1996): 357. 

9Zemach, ‘No Identification without Evaluation’; Jerrold Levinson, ‘Aesthetic Properties’, Proceedings of the Aristotelian Society 79 (2005): 191–227; see also Farrelly-Jackson, ‘Fetishism and the Identity of Art’. 

11Ibid., 223–26. 

12Jerrold Levinson, Music, Art and Metaphysics: Essays in Philosophical Aesthetics (Oxford: Oxford University Press, 2011). 

13Schmücker considered a case of computer realization of two pictures at the same time on two screens. See Reinold Schmücker, ‘Kunstwerke als intersubjektiv-instantiale Entitäten’, in Schmücker, Identität und Existenz, 161. Computer art poses special ontological problems, and one may doubt that concepts like the concept of an original are applicable in this realm without transformation of sense. 

15A more realistic, though logically distinct scenario would involve David moving back and forth between the two canvases. This version of a scenario has been suggested by an anonymous reviewer. I tend to think that the scenario does the trick. But the fact that it is more realistic comes at a price: it is unclear whether the extent to which copying is involved here relevantly changes the situation (it probably will). 

16That this link is necessary is, for example, defended in Wolfgang Künne, Abstrakte Gegenstände: Semantik und Ontologie (Frankfurt: Suhrkamp, 1983), 240–44. 

17Robert K. Merton, ‘Resistance to the Systematic Study of Multiple Discoveries in Science’, European Journal of Sociology 4 (1963): 237–82. 

18For an interesting discussion, see Darren Hudson Hick, ‘Ontology and the Challenge of Literary Appropriation’, Journal of Aesthetics and Art Criticism 71 (2013): 155–65. 

19Identifying crossbows (plural!) as original Greek or Chinese crossbows indicates that they are neither forgeries nor non-Greek or non-Chinese copies, although they may be copies from the first original Greek or Chinese crossbow. 

20This observation echoes one of the ideas Goodman discusses under the labels ‘autographic’ versus ‘allographic’ arts. Levinson would disagree, suggesting that what I describe as plagiarism or appropriation could adequately be called ‘forgery’. See Jerrold Levinson, ‘Autographic and Allographic Art Revisited’, Philosophical Studies 38 (1980): 367–83. But, even if some cases of plagiarism are covered by some concept of forgery, it seems plainly obvious that the mechanism of copying, which is involved in some forgeries but perhaps not in all forgeries, enables us to make the relevant distinction: originals that are particulars can be copied. Original inventions cannot. A fortiori, they cannot be forged by being copied. 

21There might be an additional aesthetic object – the composite object consisting of both canvases. But, from this fact, we should not conclude that the individual paintings which together compose this complex aesthetic object fail to have the status as complete originals of one work individually. 

22There is another argument developed by Rohrbaugh to the effect that artworks of various kinds cannot be identified with types, since our best conception of types is incompatible with our best conception of artworks. See Guy Rohrbaugh, ‘Artworks as Historical Individuals’, European Journal of Philosophy 11 (2003): 177–205. Rohrbaugh argues that the only viable theories of types cannot accommodate the view that paradigmatic artworks belong to this category. Others, like Ingarden, seem to suggest that our practice commits us to the existence of types, which turn out to differ radically from types as conceived of by Rohrbaugh. See Roman Ingarden, Das Literarische Kunstwerk (Tübingen: Niemeyer, 1931); see also Amie Thomasson, Fictional Objects (Oxford: Oxford University Press, 2015). The study of our conceptual apparatus concerning artworks provides valuable input for a general metaphysical theory of types. If it contradicts our best theory of types, so much the worse for our conceptual apparatus – if you are interested in something like fundamental metaphysics. But if you are after something like descriptive metaphysics, this observation may simply be beside the point. Rohrbaugh also criticizes an argument proposed by Currie about a Twin Earth scenario where two numerically distinct individuals create two aesthetically equivalent artworks (assuming that author-identity is not aesthetically relevant). See Gregory Currie, An Ontology of Art (New York: St. Martin’s Press, 1989). Rohrbaugh suggests that these artworks might change differently over time and have different modal properties and are, therefore, distinct. The details of the argument and a potential application to the case of David deserve attention. For reasons of space, I cannot go into the details here. Yet a quick comparison to the case of etching will reveal that it does not follow, either from the assumption that two originals can change over time (different prints can) or from the assumption that two originals have different modal properties, in the sense that they could have been vastly different although they are in fact similar (different prints do have these modal features), that the originals in question are originals of distinct works. This does not settle the matter. But it shows that Rohrbaugh’s considerations do not constitute a knock-down argument against the view defended here. 

24Wolterstorff, Works and Worlds of Art, 72–73. One could try to argue that there is room for making a mistake in the case of David painting Napoleon twice. A trembling hand could give rise to a mistake and, thereby, to the possibility of an incorrect instance. 

25Ibid., 73. 

28Ibid., 364. 

29Ibid., 360. 

30Ibid., 361. 

31Ibid., 360; see above. 

32We could also conceive of the situation as one where different action plans come apart in one mind – one individual action plan for each sound event. This may in fact increase chances that a critic of the argument developed here would be willing to judge the case to be a case where two improvisations are performed, rather than one performance no one in the audience can appreciate in full. On the other hand, this may also raise doubts as to the conceptual clarity of the scenario. I wish to thank an anonymous reviewer for suggesting the inclusion of the discussion of different types of action plans. 


I wish to thank several anonymous reviewers for their comments. In particular, sections on Rohrbaugh, original inventions, and the points on methodology and scope have been included in response to their nuanced concerns and extremely helpful suggestions. I also wish to thank Kai Ploemacher for going through (and thoroughly checking) the final version of the paper (all mistakes in the final version of the text remain the sole responsibility of the author, of course).

Competing Interests

The author has no competing interests to declare.


  1. Brown, Lee B. ‘Musical Works, Improvisation, and the Principle of Continuity.’ Journal of Aesthetics and Art Criticism 54 (1996): 353–69. DOI: 

  2. Currie, Gregory. An Ontology of Art. New York: St. Martin’s Press, 1989. DOI: 

  3. Farrelly-Jackson, Steven. ‘Fetishism and the Identity of Art.’ British Journal of Aesthetics 37 (1992): 138–54. DOI: 

  4. Goodman, Nelson. Languages of Art: An Approach to a Theory of Symbols. Oxford: Oxford University Press, 1969. 

  5. Hick, Darren Hudson. ‘Ontology and the Challenge of Literary Appropriation.’ Journal of Aesthetics and Art Criticism 71 (2013): 155–65. DOI: 

  6. Hopkins, Robert. ‘Aesthetics, Experience, and Discrimination.’ Journal of Aesthetics and Art Criticism 63 (2005): 119–33. DOI: 

  7. Ingarden, Roman. Das Literarische Kunstwerk. Tübingen: Niemeyer, 1931. 

  8. Künne, Wolfgang. Abstrakte Gegenstände: Semantik und Ontologie. Frankfurt: Suhrkamp, 1983. 

  9. Lamarque, Peter. ‘Work and Object.’ Proceedings of the Aristotelian Society 102 (2002): 141–62. DOI: 

  10. Levinson, Jerrold. ‘Aesthetic Properties.’ Proceedings of the Aristotelian Society 79 (2005): 191–227. DOI: 

  11. Levinson, Jerrold. ‘Autographic and Allographic Art Revisited.’ Philosophical Studies 38 (1980): 367–83. DOI: 

  12. Levinson, Jerrold. Music, Art and Metaphysics: Essays in Philosophical Aesthetics. Oxford: Oxford University Press, 2011. 

  13. Merton, Robert K. ‘Resistance to the Systematic Study of Multiple Discoveries in Science.’ European Journal of Sociology 4 (1963): 237–82. DOI: 

  14. Patzig, Günther. ‘Über den ontologischen Status von Kunstwerken.’ In Schmücker, Identität und Existenz, 107–20. 

  15. Ralls, Anthony. ‘The Uniqueness and Reproducibility of a Work of Art: A Critique of Goodman’s Theory.’ Philosophical Quarterly 22 (1972): 1–18. DOI: 

  16. Reicher, Maria E. Zur Metaphysik der Kunst: Eine logisch-ontologische Untersuchung des Werkbegriffs. Stuttgart: dbv, 1998. 

  17. Rohrbaugh, Guy. ‘Artworks as Historical Individuals.’ European Journal of Philosophy 11 (2003): 177–205. DOI: 

  18. Schmücker, Reinold. ‘Kunstwerke als intersubjektiv-instantiale Entitäten.’ In Schmücker, Identität und Existenz, 149–79. 

  19. Schmücker, Reinold, ed. Identität und Existenz: Studien zur Ontologie der Kunst. Paderborn: mentis, 2003. 

  20. Strawson, Peter. Individuals: An Essay in Descriptive Metaphysics. London: Methuen, 1959. 

  21. Thomasson, Amie. Fictional Objects. Oxford: Oxford University Press, 2015. 

  22. Wolterstorff, Nicholas. ‘Toward an Ontology of Art Works.’ Noûs 9 (1975): 115–42. DOI: 

  23. Wolterstorff, Nicholas. Works and Worlds of Art. Oxford: Oxford University Press, 1980. 

  24. Zemach, Eddy. ‘Art and Identity.’ British Journal of Aesthetics 31 (1991): 363–68. DOI: 

  25. Zemach, Eddy. ‘How Paintings Are.’ British Journal of Aesthetics 29 (1989): 65–71. DOI: 

  26. Zemach, Eddy. ‘No Identification Without Evaluation.’ British Journal of Aesthetics 26 (1986): 239–351. DOI: 

comments powered by Disqus