I argue that Dilworth has not shown the type / token theory of film identity to be non-viable, since there is no reason to think that a single object cannot be a token of two types. Even if we assume a single inheritance view of types, Dilworth's argument runs into other problems. Dilworth does not provide any convincing argument for why intentions are necessary for identifying film and why production history alone will not suffice for identifying hardly conceivable forgeries. Intention is not necessary for distinguishing between fakes and the real thing, nor is it necessary to differentiate between two artworks with the same token. Moreover, taking the notion of intentions into consideration leads to a splintering problem. I propose that production history, presentation, and non-numerical template identity suffice to identify a film on a multiple inheritance type / token theory.
film, ontology of film, types and tokens
"In Ariadne at the Movies," John Dilworth presents a promising representational account of filmic identity by arguing against the viability of the type / token theory (TTT). In this response, I propose that Dilworth's arguments do not defeat the TTT because his criticisms rely on an overly-restrictive theory of types that the TTT can coherently reject. However, even if we accept Dilworth's restricted theory of types, we should still reject his argument that directorial intention is necessary for determining film identity.
The type / token theory of film identity holds that the physical film is a token of a film type. The physical film reels are a special type of token, a template token. A template token is one from which a non-artistically relevant performance, the screening, can proceed. On the type / token theory, one can destroy a template of the film, but not the film itself. As long as at least one token remains, the film still exists.
Dilworth presents an example that purports to show that the type / token theory cannot be correct. He asks us to image a single (1) physical film made under tight budgetary constraints that necessitate a complex and compromising history of production which involves (2) a largely shared script, (3) the same cast, (4) different directors with (5) different interpretations of the story, (6) the same editor, (7) shared footage from both directors, and (8) distribution under two different titles, Greed and Sacrifice. The Greed/Sacrifice example has several complex features designed to pump our intuitions about where film identity lies, but Dilworth only explicitly bases his argument on two of the features--difference of intent (5) and different titles (8).
Dilworth thinks that Greed/Sacrifice is actually two separate films, though two films that share the same physical manifestation, i.e. the same print. If this is the case, then Dilworth thinks the type / token theory must be wrong. He argues that a token cannot be an instance of two types. To support this claim, he says that "if dogs and cows are two distinct types of animal, then there could not be a single animal that was both a dog and a cow." Though this may be true of the type dog and cow, it is not the case with types in general. A dog and a cow are also both tokens of the type "mammal:" depending on our choice of types, a single object can be a token of multiple types, just as a type can serve as a token of another type, such as dog of mammal or mammal of animal or animal of living thing. We have no reason to think that when it comes to film the situation is any different. The same object--reels of film--could be a token of two distinct types. One may object that my example involves nested types, but similar examples can be derived with types on the same level, say animals with kidneys and those with hearts. A single object, a dog or a cow, could serve as a token of each type. Dilworth's argument falls if we reject the single inheritance theory of types and tokens, but does it stand if we accept this restriction?
It is not clear if we should so readily accept Dilworth's claims that Greed and Sacrifice are two distinct films. That there are different directors at work (4) with different intentions (5) does not prove Dilworth's point, since this is the case with most films. If we bring intention into the identity of a film, then we get into all sorts of messy situations. Is Gone with the Wind David Selznick's or Victor Fleming's or George Cukor's? We do not want to say that there are three films. Given the various stages of film production and the number of different people with varying ideas about how the film should be, every film would be many in disguise. This is certainly counter to most intuitions about what counts as a film.
Dilworth does not recognize the significance of this splintering problem, as he bases his discussion on an auteurist paradigm that overestimates the role of the director in film production. It is not clear if Grand Hotel should be identified with the set designer at MGM, the producer, or with the director. I am not interested in retracing the arguments against auteur theory, but it is worth noting the problems created by Dilworth's romantic notion of film authorship.
Apart from the identity of the physical reels, what claims could we make that Greed and Sacrifice are the same film? They (2) share the same script for the most part, but this is akin to sharing a story. Hitchcock's two versions of The Man Who Knew Too Much share as much, but they are two distinct films. If we have a distinction between story and discourse, then we can see that this feature is not sufficient for sameness in film. The same cast (3) fairs no better, as any series shows. The sameness of the editor does nothing to establish sameness in film either. None of these conditions is sufficient to establish identity on its own, nor are they jointly sufficient.
Is the identity of the physical film sufficient to establish identity of the film, as Dilworth thinks the TTT assumes? A much simpler example can be concocted that would serve Dilworth just as well. Imagine that a filmmaker makes a single physical film that moves from a blue screen to a black screen to a black screen with a white dot. The film is screened under two separate titles, Drowning and Flight. If your intuitions are like mine, you will want to say that Drowning and Flight are two separate films. What we have is a Danto inspired case of perceptually indiscernible objects that are different. However, we do not need to say that they are different because they have different intentions, since the presentation can account for our intuitions. Where the title occurs might matter. If the title is on the print, then we have a different token. If we count titles, however it is presented, as part of the work of art, as Dilworth seems to do, then again we have two separate films. It seems that we could conceivably have two different titles listed on the theater awning, but the screening in the same room. Even in this case it seems that the title should still be counted as part of the work of art. Either way, the difference lies not in the history of production, but in the presentation. Viewers at different screenings or with differently title tickets would come away with radically different notions of what each phase of the film represented.
Dilworth wants to employ directorial intention into the identity condition because he thinks that "it is conceivable that there might be a numerically distinct but qualitatively identical film template, L'', that had been produced by some completely different causal process having nothing to do with Leslie's intentions and actions in directing her film. In such a case, that template L''would not count as embodying Leslie's film." The first thing we should ask is whether this is conceivable. How could another film be qualitatively identical? This would require the possibility of producing an identical replication of an allographic kind of artwork. Dilworth gives no argument as to how this would be possible. Even in my simple Drowning example, one cannot achieve qualitative identity without copying the original, which just results in another template, not another type. If we are considering forgeries, then we still have no reason to believe that qualitative identity is possible. If the work was digital, then it would be autographic and could be forged. In this case the film could just be copied. Even in the case of excellent forgeries of simple films such as Drowning, we need not rely on directorial intention to identify the original. The production history is enough. Directorial intention is just one factor of production and is not identical with directorial activity or filmmaking in general. Though we may need to resort to intentions to interpret films, there is no reason why they should be necessary for identifying them.
At times Dilworth speaks as if the film were an intention that was merely embodied. He says that "Leslie's film-making activities could have produced a numerically distinct template L', but since L' would have still been the causal outcome of Leslie's activities, it would have still counted as embodying Leslie's film Greed. It is unclear how to take this comment. What does Dilworth mean by "Leslie's film Greed" and what is it to "embody" such a thing? The claim seems to mean that Leslie's film is whatever her intentions for it to be are and that the physical film is a manifestation of this intended film. If a film is an intention, then I just made five. This cannot be what he has in mind. Perhaps Dilworth just means that L would have been Greed, Leslie's film, since she made it. This lends no support to the role of intention in identifying a particular film, since it is production history that is at issue.
Dilworth has not shown the type / token theory of film identity to be non-viable. His example is designed to show two films that share a common physical reel, a situation he thinks incompatible with the TTT. However, I argued that this situation can be accounted for by the TTT, since there is no reason to think that a single object cannot be a token of two types. Nevertheless, if we assume a single inheritance view of types, Dilworth's argument runs into other problems. There is no reason to think that Greed and Sacrifice are two different films, except for the fact that they are presented differently, under different names. I argued that Dilworth does not provide any convincing argument for why intentions are necessary for identifying film and why production history alone will not suffice for identifying hardly distinguishable forgeries. Intention is not necessary for distinguishing between fakes and the real thing, nor is it necessary to differentiate between two artworks with the same token. Taking the notion of intentions into consideration leads to a splintering problem,if we reject the overly simplistic auteurist account of film production. Rather than an unusual case, mixed intentions and dispersed control are the norm. I propose that production history, presentation, and non-numerical template identity suffice to identify a film on a multiple inheritance type / token theory.
 John Dilworth,"Ariadne at the Movies," Contemporary Aesthetics, 1 (2003), online at
 Noël Carroll develops a type token theory of film identity in "Defining the Moving Image" in Theorizing the Moving Image (New York: Cambridge UP, 1996).
University of Wisconsin-Madison