The Case Against ‘Dance Objects’

Abstract

It is generally assumed that works in the performance arts are objects. They are not physical objects, of course, but abstract objects of some kind. Such abstract objects are made available to us through the activities of composers, playwrights, choreographers and dance companies, and are appreciable through performance events. Dance objects in a strict sense, it is assumed, are such abstract objects. And, it is further assumed, such abstract objects are properly thought of as types which have particular performance events as their tokens. Dance inherits this ontological picture from classical music, where it is the central element in what I have termed the ‘classical paradigm’. I begin by briefly sketching the classical paradigm as it applies to musical works and performances, and the reasons it has been widely assumed to be ‘the only game in town’. I then present what I take to be powerful objections to the classical paradigm in its ‘pure’ form, and some desiderata for any adequate philosophical account of works in a performing art. I sketch what I take to be the most plausible strategy for defending a ‘type’ theory of works in the performing arts, Jerrold Levinson’s view of musical works as ‘initiated types’. I then argue that this view as its stands fails to account for one very significant desideratum, but might be amended so as to address this challenge. I consider here Anna Pakes’ account of dance works in her Choreography Invisible. I suggest, however, that any account of dance works as types, pure or ‘initiated’, faces serious difficulties. Finally, drawing on Wollheim’s account of multiple artworks as ‘types’, I offer an alternative account of dance works that does not commit us to works as abstract objects, and that thus denies that there are dance objects in the strict sense.

1. Introduction

Are there ‘dance objects’? In an extended sense, the answer is obviously ‘yes’, because there are objects that enter into performances of dance works. For example, in performances of Yvonne Rainer’s Room Service, it is prescribed that the dancers carry a mattress and ladders through the performance space, so, in this extended sense, the mattress and the ladders are dance objects. We might generate some interesting philosophical questions about dance objects even in this extended sense, parallel to the questions we might raise about the status of ‘props’ used in theatrical and film performances. For example, the two ‘pistols’ that appear on stage in a performance of Hedda Gabler represent real guns, but are not (one hopes) themselves real guns. In the case of Room Service, we might ask whether, in the context of a performance, the object manipulated is a mattress or represents a mattress. In general, how we answer this question seems to depend upon which objects would be acceptable replacements for the object used in the performance. If all that is required is that the object manipulated by the dancers has the appearance of a mattress, then whatever object is used—even if it actually is a mattress—represents a mattress in the context of the dance performance. The same presumably applies to the ‘pistols’ in a performance of Hedda Gabler. Arguably, if the point of Rainer’s work is to get the audience to focus on the lived intelligence of the human body in its performance of everyday tasks (Carroll and Banes (1982), responding to Beardsley (1982)), it is an actual mattress that is required in this case, whereas in another work anything appearing to be a mattress would do.

Another sense in which there may be ‘dance objects’ is if we take the entities that execute the movements conceived by a choreographer as ‘objects’. This might sound offensive if these ‘objects’ are also subjects—human dancers whose execution of various movements reflects their training and interpretive expertise. But presumably no such offense would arise if the movements are executed by objects that have been programmed to do so, as in recent works involving AI robotic ‘dancers’.

While these issues may be interesting, they are not the ones that concern me here. What I want to address is the objecthood of the dance work itself, taken to be a focus of aesthetic and artistic appreciation. That an object serves as the focus of our appreciative perceptual attention to an artwork is a familiar idea in the reflective self-understanding of other art forms, such as sculpture and painting, and it is usually assumed that this object is itself the work of art. The ‘object’, in this case, is a physical entity that, at any given time, has a particular physical location, that is brought into existence by the actions of one or more individuals, and that may cease to exist as a work if it ceases to possess certain properties taken to be essential to its functioning as a particular focus of artistic appreciation. In other art forms, the immediate focus of our appreciative perceptual attention to an artwork is not an object but an event. It is an event—a screening—to which we attend in our attempts to artistically appreciate a film, and it is also an event—a performance—to which we attend in our attempts to artistically appreciate a work in the performing arts—a musical work, a theatrical work, or a work of dance. But in these cases, it is assumed that the work we are seeking to appreciate is accessible through, but only rarely identical to, one of these events. It is Casablanca, Sibelius’s Second Symphony, King Lear, or Room Service that we seek to appreciate on such occasions. In the case of the performing arts, our appreciative interest may also be in the event itself, since we assume that different performance events may reveal different qualities of a performed work, and may do so as a result of the creative interpretation of the work by the performers. Arguably, in such cases, the performance event may itself be a work of art.1

These considerations bear on our theme—the existence and nature of dance objects in a non-extended (or, as we might say, strict sense)—because it is generally assumed that, even though works in the performance arts like the ones just cited are made accessible to us through performance events, the works themselves are objects. They are not physical objects, of course, but abstract objects of some kind. Such abstract objects are made available to us through the activities of composers, playwrights, choreographers and dance companies, and are appreciable through the performance events described above. Dance objects in a strict sense, it is assumed, are such abstract objects. And, it is further assumed, such abstract objects are properly thought of as types which have particular performance events as their tokens. Dance inherits this ontological picture from classical music, where it is the central element in what I have termed (D. Davies 2011, chapter 2) the ‘classical paradigm’.

I have argued recently (2024) against the idea that those artworks that can be properly appreciated through encounters with different objects or events—what have been termed ‘multiple artworks’—should be viewed as ‘types’ in the sense understood by proponents of the classical paradigm. In the rest of this paper I shall argue, more specifically, for the view that dance works should not be construed as types in this sense. If, as suggested above, there are dance objects in a strict sense only if dance works are abstract objects, then my conclusion will be that there are no dance objects in a strict sense. But, I shall further argue, this should not concern us since we can give an adequate and more plausible account of the nature and appreciation of dance that eschews any such objects.

The overall structure of my paper is as follows. I begin by briefly sketching the classical paradigm as it applies to musical works and performances, and the reasons why it has been widely accepted. I then present what I take to be powerful objections to the classical paradigm in its ‘pure’ form, and some desiderata for any adequate philosophical account of works in a performing art. I sketch what I take to be the most plausible strategy for defending a ‘type’ theory of works in the performing arts, Jerrold Levinson’s view of musical works as ‘initiated types’. I then argue that this view, as it stands, fails to account for one very significant desideratum, but might be amended so as to address this challenge. I consider here Anna Pakes’ account of dance works in her Choreography Invisible. I suggest, however, that any account of dance works as types, pure or ‘initiated’, faces serious difficulties. Finally, drawing on details of Wollheim’s discussion of multiple artworks as ‘types’, I offer an alternative account of dance works that does not commit us to works as abstract objects, and that thus denies that there are dance objects in a strict sense.

2. The ‘Classical Paradigm’

The ‘classical paradigm’ is a very widely accepted view in the analytical philosophical tradition that purports to explain the nature and the appreciability of the kinds of performable works of art cited earlier—for example, Sibelius’s Second Symphony, Ibsen’s Hedda Gabler, and Balanchine’s Agon. Those philosophers who played the most significant roles in developing the classical paradigm—Wollheim ([1968] 1980), Kivy (1983), Wolterstorff (1975, 1980), Levinson (1980), S. Davies (2001), and Dodd (2007)—generally took classical music as a model for performable works more generally—hence the name of the paradigm. They took the primary explananda for a philosophical theory of performable works to be their repeatability and their appreciability. The repeatability of a performable work consists of the fact that different performance-events can be fully qualified to provide experiential access to the work. But we cannot identify the work with those performance-events, individually or collectively, because (a) there can be performable works that are never actually performed, and (b) any given performable work could have had a different number of performances. Such works, therefore, must in some sense exist independently of their performances. But, since, as noted earlier, the experientially appreciable properties of performable works are given through their performances, we also need to explain how the work itself can possess these properties. Consider, for example, a work like Sibelius’s Second Symphony, of which one critic said the following: ‘The pastoral quality of the opening Allegretto is tinged with melancholy, and there is a solemnity in the triumph of the work’s conclusion’ (Day (1991)). While one can easily imagine how a performance of this work might have these qualities, what are we saying when we ascribe such qualities to the work?

The canonical response to this challenge by philosophers writing in the analytic tradition is that, as noted above, musical works, and, it is assumed, performable works more generally, are types that have performance-events as their tokens. Types, it is widely assumed, are abstract objects that by their very nature cannot possess the experienceable properties of their tokens, because abstract objects, having no determinate spatio-temporal location, do not possess properties that depend upon occupying such a location and causally interacting with experiencers. But we can correctly apply the same predicates to works that we apply to their tokens, albeit with a different sense, and this explains what we are doing when we describe Sibelius’s Second Symphony as ‘tinged with melancholy’.

Let me unpack and historically situate these ideas to give some flesh to this skeletal characterization of the classical paradigm. In the contemporary analytic tradition, the idea that performable works, and multiple artworks more generally, are types appreciable through those instances that are their tokens can be traced to the first edition (1968) of Richard Wollheim’s Art and its objects. But it has also been assumed that Wollheim’s account needs to be modified in certain ways to meet more general metaphysical requirements. Two such modifications were first proposed by Wolterstorff (1975, 1980). First, because performable musical works admit of performances that may lack some of their prescribed musical features—a performance of a work may possess one or more wrong notes, for example—they are best viewed as ‘norm-kinds’ or ‘norm-types’, kinds or types whose instances are correctly formed if they possess certain properties. Wolterstorff argues that natural kinds are also norm-kinds. For example, while the capacity to growl is rightly predicated of the type ‘Grizzly Bear’, a creature lacking this capacity as a result of a birth defect is still a grizzly bear if it has the appropriate genetic history. Second, while Wollheim held that types share with their tokens those properties taken to be definitive of the type, Wolterstorff claimed that what are shared are predicates that are used analogically when applied to the type: while ‘growls’ is correctly predicated of both the norm-type Grizzly Bear and most tokens of that type, what is thereby predicated in the former case is the property of having well-formed tokens that growl. Similarly, while ‘opens on a C#’ may be correctly predicated of both a musical work and most performances of that work, what is thereby predicated in the former case is that well-formed performances of the work have that experienceable quality.

Both of these modifications of what is taken to have been the Wollheimian view are preserved by Dodd (2007), but he maintains that, once we clarify the metaphysics of types, the idea that performable works are norm-types has consequences that run counter to widely shared assumptions about such work (these consequences were brought out by Kivy (1983) in an earlier defense of what he termed ‘musical Platonism’). Drawing upon work in theoretical metaphysics, Dodd maintains that a type is an abstract object that exists eternally if it exists at all and is modally and temporally inflexible. So, if performable works are norm-types, they must be discovered, rather than created, by their ‘composers’. Furthermore, according to Dodd, such works are individuated only by reference to the properties they possess eternally as sound-structures. So, if Sibelius’s Second Symphony is a norm-type, it is a pure sound structure, discovered by Sibelius, whose constitutive properties are all and only its sonic properties (or, for Dodd, its timbral sonic properties). That Sibelius discovered this structure in a given musico-historical context that included earlier and later works by himself, and others, is an interesting (and appreciable) fact about Sibelius, but this has no bearing on the appreciable properties of the musical work qua type. And, if we take the classical paradigm to also apply to performable works of theatre and dance, then Shakespeare, Pirandello, Balanchine and Yvonne Rainer were also discoverers rather than creators of their works, and the proper appreciation of those works must prescind from their contexts of creation. Agon, as discovered by Balanchine, may reflect more general facts about dance practice and views about dance at the time of its discovery, but this will not bear on our correctly appreciating the dance work Agon if, as a follower of Dodd might maintain, it is pure structure of action-types. Let me term types as understood in the way that Dodd proposes—entities that are eternally existing, uncreatable, abstract objects—‘D-types’. If performable works are D-types, then as noted above, they are eternally existing sound structures that are discovered rather than created, and that lack any properties that derive from their context of discovery.

But ‘one philosopher’s modus ponens is another philosopher’s modus tollens’.2 Levinson (1980), responding to Kivy’s original defense of musical Platonism, took being created, and being individuated by properties deriving from their musico-historical contexts of creation, to be defining conditions of musical works. As applied to a dance work like Agon, this means that being brought into existence in a particular dance-historical context is partly constitutive of the work, and to properly appreciate it as the dance work it is requires that we take account of those properties that the prescribed movements possess in virtue of these historical facts. For example, Balanchine’s incorporation of unclassical elements in the piece, reflecting his interest in black dance culture and black dance movements, raises the question, disputed in critical responses to the work (see, e.g., Tomiḉ-Vajagić 2020), whether this served purely formalist purposes or was a response to the social struggle at the time between the Africanist and the dominant white cultures.

Multiply performable works of music, theatre, and dance, therefore, cannot be identified with the eternally existing and unchanging norm-types proposed by Kivy, Wolterstorff, and Dodd—Levinson terms these implicit types. Implicit types of sound or sound/performance-means structures are implicit in the space of possible sounds, just as various legal games are implicit in the rules of chess. Rather, according to Levinson, multiply performable works are initiated types that come into existence through acts of indicating implicit types performed by particular individuals in particular art-historical contexts. A musical work, then, is a sound/performance-means-structure-as-indicated-by-a-given-individual-X-at-a-given-time-T, or, alternatively, a sound/performance-means-structure-as-indicated-in-a-given-musico-historical-context-C. Such entities, Levinson claims, can satisfy the desiderata of concern to the Platonist. A dance work, as an initiated type, is repeatable—it admits of multiple right performances—and can analogically share the experienceable qualities of its tokens. But initiated types also respect our widely held beliefs that performable artworks, like other artworks, are created by artists and possess appreciable properties that are grounded in their art-historical contexts of indication.

Performable works as initiated types do not, it is true, satisfy certain other desiderata proposed by some philosophers who oppose the identification of performable works with types of any sort. In particular, contrary to the claims of Rohrbaugh (2003), they are not modally and temporally flexible. According to Rohrbaugh, the very same performable work could have differed in what it requires of its correct performances, and these requirements might themselves change over time, whereas neither of these things can hold of a type, whether implicit or initiated. But it is difficult to find compelling arguments for this view, and, failing such arguments, it seems that the type theorist of either stripe can explain apparent instances of modal and temporal flexibility by positing that we have very similar but different works that possess the relevant properties. When a composer of a performable work, or the members of a performing company, returns to an existing work and changes some of its structural properties, the type theorist will claim that we have a distinct but closely related work. As evidence for this analysis of the situation, she might point to the fact that the original requirements can still be met in a performance even after the composer makes such changes.

There is, however, one further desideratum—the ‘problem of variability’ (D. Davies 2012)—that poses a problem for the type theorist who identifies a performable work with a norm type, whether implicit or initiated. Unlike literary works, where every well-formed instance must possess all of those linguistic properties that bear upon a work’s proper appreciation, well-formed performances of performable works can differ in such properties. This is one reason why aficionados of a particular performable work may seek out different performances of that work, whereas aficionados of a literary work do not seek out different copies of the same novel. This does not require that one subscribe to the idea that the requirements for correct performance of a performable work may change over time. Rather, the differences between performances of a performable work that constitute its variability in the relevant sense are differences within the ‘envelope’ of correct performances allowed by a single set of requirements in place at a given time.

The problem of variability does not depend upon a particular understanding of what makes a performance ‘well-formed’ or ‘correct’, or ‘flawed’ or ‘incorrect’. How we should understand the normativity to which such evaluations appeal will vary according to a theorist’s more general philosophical orientation. It may be located, for example, in a norm that transcends our practice, as in the case of Dodd’s ‘norm-types’, or in a norm internal to the presentational practices in the art-historical context of a work’s initiation, as a proponent of historically ‘authentic’ performance might claim, or in a norm that itself changes as the presentational practices in a given art form change, as a proponent of the modal and temporal flexibility of performable works (e.g., Rohrbaugh) might claim.

But the problem of variability will arise for any theorist who is committed to the following claims:

(1)

At least some performances in the performing arts are performances of something else, a ‘work’.

(2)

Such performances are treated, by those involved as initiators of works, presenters of performances of those works, and attendees at those performances, as falling under norms that allow comparative assessments to be made as to a performance’s fidelity to the work.

(3)

Those instances of works that are classified as ‘well-formed’ in this sense can vary in the manifest properties that bear upon the appreciation of the work.

Some philosophers do not face the problem of variability because they reject (1). James Hamilton’s account (Hamilton 2007) of theatre, for example, holds that theatrical performances are never of theatrical works: scripts originating in the writings of a playwright are merely one ‘ingredient’ that a performing company may take into account. A similar view of musical performance has been defended by the Ghent-based group MusicExperiment21 (2015). And Goehr (2007) has argued persuasively that the idea of musical performances as governed by the ‘work concept’—as aiming to be ‘true to the work’—only obtained purchase upon musical practice in the mid-18th century. But since my concern in this paper is to challenge the idea that there are dance works and that they provide us with ‘dance objects’, I shall assume that those whose views are under consideration accept ‘1’. I also assume that, whatever their other differences, such thinkers subscribe to (2) and (3). ‘(2)’ is built into the idea that, if artworks that admit of multiple instances are kinds, they are ‘norm-kinds’, and on this at least Dodd, Levinson, Pakes, and Rohrbaugh would agree. And ‘(3)’ is the distinction, just noted, between the role that work-instances play in the appreciation of works of literature, and their role in the appreciation of performable works of music, theatre, or dance.

The challenge posed by the variability of performable works, then, is to explain the range of variations that is permitted in correct performances of a given performable work when we hold constant its requirements.

It might be thought, however, that such variability poses no threat to the type theorist who identifies a performable work with a norm-type, since the very norm in terms of which a given norm-type is individuated itself allows for a range of different interpretative performances that have the properties that the norm dictates. But, as Wolterstorff himself noted (1975), composers of performable works leave many of the constitutive features of the work unspecified, relying on the understandings of anticipated performers both for things left completely indeterminate in the score, stage directions, or choreographic details and for things that, while specified, require interpretation. Right performances of the work are ones that interpret what is specified correctly, and that supplement this interpretation by supplying the understandings assumed by the composer but left unspecified. Nor could the problem of variability be solved by requiring that composers of performable works make explicit all of the normative requirements for their works, even if this were possible, since explicit norms themselves require interpretation—this is the familiar Wittgensteinian point that requiring a rule for the correct application of a rule generates an infinite regress. To solve this problem, we must appeal in some way to the practice of a performative community3. But for the Platonist, whose norm-types are supposed to exist timelessly outside of human practices, this problem seems insuperable: if the range of variability of a performable work is inseparable from the practices of an interpretative community, it cannot be accounted for by something that lies outside those practices.4

Can the indicated structure theorist do any better? For Levinson, the performable work of music, as an initiated type, is partly constituted by salient features of the musico-historical context in which the work is composed, but he retains the Platonistic idea that the norms governing right performance are embedded in the implicit types that are indicated in a given context. But if, with Levinson, we are prepared to include in our ontology entities that are partly constituted by such historically situated entities, could we not also include entities that are partly constituted by the particular presentational practices in place in the context of composition—that is, by the practices that, I have suggested, determine how the strict instances of a performable work may vary? We find such a view of performable dance works in Anna Pakes’ Choreography Invisible. Drawing on Levinson’s notion of ‘initiated types’, she suggests that performable works of dance are properly understood to be ‘structures of action-types indicated by choreographers within a given historico-artistic context’ (Pakes 2020, p. 181). In chapter 9, however, she considers the problems posed for such a view by the ‘variability’ of dance works. In addressing such problems, she suggests that Levinson’s notion of an ‘indicated structure’ might be modified so as to include the relevant presentational practices in place in the context of indication. Such an entity is an initiated type in Levinson’s sense that also incorporates those norms implicit in dance performance practice that explain the range of variability in right performances of a work. A choreographer, or a company, initiates a dance work by prescribing tokens of a sequence of action types, and something counts as a strict instance of the work if it presents a sequence of action tokens that would be judged within the relevant performative community to be a right instantiation of what was prescribed. The problem of variability can be answered by such a modified version of the ‘indicated structure’ account since the norm-embodying practices determining a performable work’s range of variability are built into the work itself. In this way, we can resolve a major difficulty facing a type-theory of performable artworks. Dance works, on Pakes’ account, are initiated types of action-structures, and these types, as abstract entities, provide us with dance objects in the strict sense5.

But, once again, ‘one philosopher’s modus ponens is another philosopher’s modus tollens’.6 Dodd’s response to Levinson (Dodd, 2000, 2007) is that, if the only way to reconcile repeatable works appreciable through their performances with the requirements of creatability and contextualism is to posit ‘initiated types’, then so much the worse for these requirements. For, Dodd claims, by their very nature, types, if they exist, exist eternally, and this will also apply to ‘initiated types’ if they are indeed types. Both Levinson and Pakes recognize the force of this objection. Levinson’s considered response is expressed in a 2012 paper (Levinson 2012, pp. 56–57) where he reconsiders his view of repeatable musical works. If performable works are indeed what we have termed D-types, then, Levinson acknowledges, he must give up on the idea that the ‘indicated structures’ with which he identifies such works are types. If types are eternal and uncreatable, ‘that is precisely the opposite of what indicated structures and initiated types are supposed to be, and so insisting that they are types would undermine one of the main motivations for introducing them’. But in canvassing alternative understandings of indicated structures, Levinson reveals a residual commitment to the Platonistic idea of implicit types to which a composer does something: this is retained even if the idea of musical works as themselves types is repudiated. Indicated structures cannot be ‘qua-objects’ like ‘Obama-as-President’, he maintains, because such things are ‘too insubstantial, too aspectual to be things we create in any robust sense by doing something to a preexisting sound structure’, i.e., an ‘implicit type’. Artistic indication, he further claims (Levinson 2012, p. 52), involves the following: (1) there is a deliberate process of selection and choice; (2) it appropriates the sound-type structure as something associated with the artist in question; (3) it implies an attitude or process of approval, that is some level of acceptance that this is the appropriate sound-type structure for the circumstances (maybe to address a particular artistic problem); and (4) it involves establishing ‘a rule, a norm, a miniature practice, whereby pianists play a piece by Chopin and not just any piece of music when they play that sequence of notes chosen by Chopin, and do so precisely because that sequence was chosen by Chopin’ (ibid., p. 54).

But how, then, should we understand the product of an act of indicating an implicit type?

Levinson says very little about this, suggesting only that indicated structures are ‘what Wollheim called generic entities, that is, things that can have instances’. But, as demonstrated in our review of the ‘classical paradigm’, Wollheim’s account of ‘generic entities’ has been taken to be a principal source, albeit with a need for metaphysical precisification, of the view that musical works are D-types! So how can an appeal to Wollheim aid Levinson in his attempts to rescue the ‘indicated structure’ view from Dodd’s objections?

The same worries might be raised concerning Pakes’s proposal, as she acknowledges in the later chapters of her book. First, if indicated structures of action-tokens are repeatable and contextualised, what are we including in our ontology when we allow it to encompass abstract entities whose constitutive elements include particular events, individuals, and practices? Dodd’s response to Levinson heightens this problem. Even if we were prepared to acknowledge such entities, it seems they must be D-types and thus uncreatable and unchangeable, thus defeating the very purpose of positing such ‘indicated structures’ in the first place. And, it might also be asked, how are we able to explain repeatability and appreciability without D-types?

But Pakes also brings out, in her discussion, problems that arise if we try to apply to dance the idea, central to Levinson’s account of musical works, that works are indicatings of implicit types. This idea owes its plausibility to the thought that, if we can delimit in a principled way a set of possibilities realizable within some system of constitutive principles, then it makes sense to say that these possibilities are implicit in those principles and therefore already exist to be discovered. As noted above, the chess analogy is often offered here. If, like Dodd, we take musical works to be sound structures that are individuated in terms of their timbral sonic properties, then performable works of classical music might be thought to be implicit in the organizing principles of the sonic space containing such sound structures. This idea is far from unproblematic, but at least in the musical case it has enough plausibility to attract philosophers like Dodd and Levinson, whose disagreement rests not on the existence of implicit musical types but on whether musical works can be identified with such types.

But in the case of the contextualised structures of action-types with which Pakes proposes to identify performable works of dance, it is difficult to identify the principles constitutive of the space of possible agency in which such structures might be thought to exist implicitly. The seriousness of this difficulty is perhaps masked by the initial example offered by Pakes of such structures. In classical ballet we have a vocabulary that distinguishes the types of prescribed movement out of which particular dance works might be composed, and these works might then be seen as implicit in a possibility space defined in terms of those types of movements. As Pakes says, ‘the range of action- types employed in dance is very large. It includes (but is not limited to) steps from a codified vocabulary handed down by a tradition, as in the case of the five steps of the basse danse, combined into different sequences to constitute difference dances… Similarly, ballet’s academic lexicon consists of a series of action- types— plié, assemblé, chassé, arabesque, entrechat, and so on— which include movements and poses.’ (122) Even here, as Pakes notes, there are problems, given historical differences, evident in photographic documentation of earlier performances, in the ways such action-types are to be executed. These changes reflect the evolution of norms, grounded in performance practice, as to how basic action-types are to be rendered in bodily movements.

But once we move away from classical ballet, the action-types out of which a particular dance work might be created seem themselves to be created by choreographers and dancers in the act of initiating a dance work. This issue comes up in Pakes’s discussion of dance preservation (213–4):

‘Dancers are thereby treated as sources of insight into the overarching structure and its purposes, as well as the detail of specific action- types. Even a choreographer who is still present and involved in restaging may struggle to recall the latter, if she has not danced in the work herself. This is especially the case for those action-types devised or drawn from original dancers’ individual capabilities and distinctive habitus… In the process of devising and rehearsal, action-types and their organisation become ingrained in dancers as a set of physical capacities or embodied knowledge-how. And somatic memory may therefore be more effective than recall in imagination or reflective consciousness.’

This problem is further exacerbated by the fact that, whereas there is a shared understanding in the post 18th C classical musical tradition of how a given sound structure is to be formally characterized, there is no agreement on which features of a dancer’s movement are to be notated. As Pakes documents, notational schemes differ as to which features of a given bodily movement define that movement as an element in a dance, with consequent implications as to which other movements count as executions of the same dance element. Furthermore, notations of a dance work usually record a particular performance rather than preexisting and guiding it. It is this fact that partly underwrites the reliance, noted above, on a dancer’s body memory as the means whereby a dance work is preserved.

So, to summarize, even if we can make sense of the idea that a composer of a classical musical work indicates an implicit type, it is unclear, in the case of most dance works, how the possibility-space grounding talk of preexisting ‘implicit types’ is to be characterised, and even more apparent that, if we identify a dance work with a structure of action-types, the structure is frequently created rather than discovered. In other words, not only are dance works not essentially indicated but also their indication cannot in general involve implicit types.

3. Dance Works as Wollheimian Types

Levinson, as we saw, appeals to Wollheim’s idea of generic entities in trying to address these problems, and, as we also noted earlier, it is generally accepted that the generic entities with which Wollheim proposes that we identify performance artworks are types and, if Dodd’s claims about types are accepted, are properly viewed as D-types. But I think the answer to Levinson’s and Pakes’s concerns does indeed lie in Wollheim’s discussion of generic entities if we divorce this discussion from subsequent debates over the metaphysical status of types. We can do this if we look more closely at what Wollheim says when he introduces the idea of multiple artworks as ‘types’.

As we saw earlier, Wollheim has been taken by writers such as Wolterstorff and Dodd to be a progenitor of the view that multiple artworks are types, where the latter are understood to be abstract entities. But, as I have argued elsewhere (D. Davies 2024, chapter 9), this reading cannot survive a careful reading of Wollheim’s text. In considering how we should view multiple artworks (Wollheim [1968] 1980, pp. 34–35), he distinguishes between two questions we might ask:

(1)

The ‘logical question’: ‘Are works of art physical objects?’ The options here are to answer positively and say that works of art indeed have the logical status of particulars or to answer negatively and identify works of art with non-particulars—for example, with universals or classes.

(2)

The ‘metaphysical question’: ‘Are works of art physical objects?’ Here, assuming a positive answer to the logical question, we weigh alternative accounts of the kinds of particulars they are—for example, physical particulars or, as in the case of what Wollheim terms the ‘Croce-Collingwood’ view, mental particulars.

The claim that multiple artworks are types is presented by Wollheim as one way in which we might answer the logical question negatively; to classify multiple artworks as types is to claim that they are not objects or particulars of any kind, physical or abstract. Types, for Wollheim, are like classes and universals in being generic entities, ways in which particulars are grouped, and what distinguishes different kinds of generic entities is how they group such particulars. Types, he maintains, differ from classes and universals in being more closely related to the particulars falling under them. Indeed, they are sometimes treated as if they were themselves particulars, something manifest in our predicating of a type the same properties that we predicate of its tokens.

This explains why, for Wollheim, types and their tokens share properties, rather than, as Wolterstorff claimed, only sharing predicates. It is because Wolterstorff takes performable works to be abstract particulars that he requires that experiential predicates be used analogically when applied to them. Wollheim’s view of such matters, however, is better captured by Stefano Predelli’s interpretation of generic expressions (Predelli 2011), where predicates are applied univocally while it is expressions like ‘Agon’ that can be used equivocally to refer to both a dance work and a particular performance of that work. In the former case, according to Predelli, it is used to talk collectively about correct performances. Thus, for Wollheim, there is no problem explaining how a work can correctly be ascribed a property like being ‘tinged with melancholy’. Furthermore, turning to Wolterstorff’s other ‘correction’ of Wollheim, performable works are not norm-types because they are not types at all in the metaphysical sense—the norms operative in sorting performances are embedded in the relevant performative practices.

While it is part of our linguistic practice to treat what we may term Wollheimian types in the same way that we treat those particulars that are their instances, this does not support the idea that such types are themselves particulars—albeit abstract—that exist independently of their tokens. It merely reflects, Wollheim claims, that types are things characteristically postulated ‘where we can correlate a class of particulars with a piece of human invention; these particulars may then be regarded as tokens of a certain type’ (Wollheim [1968] 1980, p. 78). He takes this to be an answer to an ‘entirely conceptual [question] about the structure of our language’. Types are individuated by reference to the generative relationships in which particular groupings of tokens stand to such pieces of human invention—they are identified in terms of ‘how the tokens of a given type are generated’ (1980, p. 79), or how ‘the tokens can be generated from the initial piece of invention’ (1980, p. 80). In the case of multiple artifacts, Wollheim distinguishes three ways in which an act of human invention can establish the conditions under which multiple correct instances of a given artifact can be generated. First, an inventor can produce an instance of the artifact which serves as an exemplar, where conventions in place in the receiving community determine which properties of this exemplar must be reproduced in other correct instances of the artefact. In the arts, this model applies to works of literature. Second, she can produce what Wolterstorff (1980, p. 90ff) terms a ‘production artifact’ which, when appropriately used given conventions in place, generates correct instances of the multiple artifact. This model applies to the production of coins in a mint and, in the arts, to works of cast sculpture (where the production artifact is a moulding), prints (where it is a matrix), and photographic and cinematic works (where it is a negative, a bit map, or a master encoding). Finally, as in the case of recipes and multiply performable works of art, the inventor provides directions for performing a sequence of actions which, when executed in accordance with the relevant conventions in the performing community, issue in correct instances.

In each of these cases, multiplicity or repeatability is built into the contextualized act of initiation performed by the artist. This does not obtain with a singular artwork like a painting, however, even though a perfect copy of a painting might provide us with the same visual experience as the original. The experientially appreciable properties of the work are those possessed by its correct instances, those that result from applying the relevant conventions to the product of the artist’s creative act. Where a multiple artwork is variable, in the sense defined above, the range of variability is again determined by conventions governing the generation of instances implicit in the practices of the relevant presentational community. This is how the Wollheimian account speaks to the problem of variability. And we have also seen how it explains the two features of multiple artworks that have been taken to motivate Platonistic approaches—repeatability and appreciability.

4. The Ontological Status of Wollheimian Types

The foregoing analysis of generic entities, what I am terming Wollheimian types, offers a way in which a defender of the idea that performable artworks are ‘indicated structures’ might, as Levinson suggests, appeal to Wollheim’s account of generic entities in distancing this idea from the idea that such works are D-types. But, if Wollheimian types are not D-types, what should we say about their ontological status? If, as Wollheim claims, we postulate a Wollheimian type when we correlate a group of particulars relative to a piece of human invention, what is it, ontologically speaking, that we postulate? We see here the temptation to answer this question by appealing to something transcending our actual practices that justifies the groupings of particulars that we make. But, as we have seen, what justifies these groupings, for Wollheim, is norms embedded in our presentational practices; no further external justification is needed. Furthermore, Wollheim insists that this ‘internal’ justification exemplifies a much more general practice in our thinking about multiple artifacts. His initial examples of ‘types’ include the Boeing 747 and coins from a mint, but a further example, in this spirit, might be the current Special Issue of Arts on dance objects! In each such case of multiplicity, one or more individuals, acting in the context of a practice, initiate a multiple artifact by bringing into existence a means for generating multiple instances that meet standards implicit in that practice. If we need to posit a D-type transcending and preexisting our practices in order to make sense of those practices in the case of Agon, why would this not be necessary in the case of the Special Issue of Arts?

Levinson and Pakes, however, if they appeal to Wollheim’s account of generic entities to make sense of the idea of ‘indicated structures’ that are not D-types, might insist that, in order to be the bearers of the properties that we ascribe to multiple artworks, a Wollheimian type must be an entity of some sort, and, more specifically, the product of an act of initiating a given multiple work. But how is this to be understood if it is not a D-type? Suppose, for example, it is claimed that the ‘work’ or ‘artifact’ type in the case of a performable work is a set of contextualized instructions generated by the initiator. But this cannot be a particular textual inscription of those instructions: if Sibelius had used a different score to inscribe his instructions for the Second Symphony, this would not have been a different work! But if we speak instead of the content of what is inscribed, this might seem to reinstate D-types through the back door! In this way of adapting the idea of Wollheimian types to the needs of the ‘indicated structure’ theorist, we would indeed have dance objects, but those objects would seem to be eternal and uncreatable.

I suggest an alternative response to the present challenge. Consider one of Dodd’s original criticisms of Levinson’s claim that indicated structures are initiated types. The temporal elements built into the idea of an indicated type, Dodd suggests, make them resemble events rather than objects (Dodd 2000, p. 440). While this is intended as a reductio, it might actually provide a better understanding of what Wollheimian types are; they are token actions that license a multiplicity of instances by providing a certified exemplar, a production artifact, or a set of instructions that, when utilized according to conventions in place, issue in experientially appreciable instances of the work. A work’s experientially appreciable properties are properties experienced through these instances—the latter present the ‘focus of appreciation’ specified by the artist—and in talking of ‘the work’ in Predellian terms we are talking of its correct instances. But these are also properties correctly ascribed to the work viewed as a contextualized act of initiation because the latter is the contextualized specification of a focus of appreciation having these properties. To say that the first movement of Sibelius’s Second Symphony is ‘tinged with melancholy’ is to say that Sibelius’s contextualized action was an initiation of a sound structure whose right instances have this property. I have defended the idea of artworks as generative performances elsewhere (D. Davies 2004, summarized in D. Davies 2024, chapter 9). My claim here is just that if, as Levinson and Pakes maintain, performable works are both creatable and contextualized, and if they therefore cannot be D-types, the best ontological account of such works is that they are Wollheimian types understood as contextualized indicatings, contextualized initiatings that license foci of appreciation in the sense just sketched. But if so, then, to return finally to our theme, dance works are not ‘dance objects’ in the strict sense.