In Part One I argue that we should consider fictional characters not to be the strange occupants of another realm but rather as abstract cultural artifacts as ordinary as the works of literature in which they appear. If we do so, then the problems that arise for fictional characters, including how to refer to them and offer identity conditions for them, parallel problems for other abstract entities and cultural artifacts. I have attempted to show how these problems may be overcome by considering fictional characters to be dependent entities, thus removing two major stumbling blocks to postulating fictional characters and suggesting ways to overcome these problems for other dependent abstracta. But so far this merely eliminates reasons not to postulate fictional characters and does not provide us with any positive reason to postulate them.
It is now time to face the second question: Should we admit fictional characters into our ontology? The case of fictional characters provides occasion for reexamining the larger issue: What, in general, should we admit to our ontology, and on what grounds? One of the points of philosophy in general, and ontology more particularly, is to help us make sense of our world. Thus one goal in choosing an ontology is to select one that provides an adequate basis for understanding our experience of and discourse about the world. Naturally this does not mean that we can never determine that experiences are misleading, discourse fallacious, or practice in need of revision. It only requires that we seek a theory able to analyze what our experience is about and whether our sentences are true or false as consistently, adequately, and elegantly as possible overall.
Although examples from fiction and mythology have long provided a source of interesting puzzles and counterexamples that have guided the development of theories from Frege to Russell to Kripke, fiction has been seen as a sideshow issue in metaphysics. Even the Meinongian minority, which has done much to bring the topic of fiction back into discussion, has done little to dispel the image of fiction as a strange metaphysical jungle beyond the boundaries of traditional metaphysics.
Lying behind the sideshow view of fiction is an assumption shared by believers and disbelievers in fictional objects alike: Fictional characters are (if anything) odd, freakish entities, quite unlike common or garden objects. Disbelievers have used the supposed freakish nature of fictional entities as grounds for rejecting them, alleging that they would be too unruly to accommodate in a theory and fearing that by handling such oddities we will be led into contradiction. Believers have boldly, smilingly embraced their odd creatures, proposed special ontological realms to house them, and shown how, by handling them carefully, we can accommodate their curious tendencies and avoid contamination by contradiction.
The key to seeing the centrality of fiction in metaphysics lies in giving up this assumption and recognizing the similarities between fictional objects and other entities. In the view I propose here, fictional characters are abstract artifacts – relevantly similar to entities as ordinary as theories, laws, governments, and literary works, and tethered to the everyday world around us by dependencies on books, readers, and authors.
Dominant theories of the reference of names have emphasized that names, unlike descriptions, function by means of a direct reference to their objects, and that causal and historical circumstances play an essential role in our ability to refer to objects by name. This model seems to break down in the case of fictional names: If fictional objects are not spatiotemporally located, then it seems they must also be causally inert, making it inconceivable how causal or historical circumstance could play any role in the reference of fictional names.
This supposed incompatibility between the claim that fictional names refer and the claim that causal or historical features are essential to the reference of names has provoked many to reject the thesis that fictional names refer. Traditional causal theories of reference treat names of fictional characters as nonreferring terms, and in Naming and Necessity Kripke goes still further to argue that genuinely fictional names cannot refer to any actual or possible object. Indeed, anyone who wants an even partially naturalistic account of the reference of names, taking causal or historical chains to play an essential role in the reference of names, has cause to worry about the case of fictional names. Those who take fictional discourse seriously have on occasion taken the other horn of the dilemma, maintaining that because causal or historical theories cannot allow that fictional names refer we should abandon an across-the-board causal-historical model of reference.
Troubles with reference present difficulties for those postulating fictional objects, for if we cannot successfully refer to fictional characters at all, then there seems little point in postulating them, and the process of acquiring knowledge about them becomes mysterious.
I have argued that we can offer better analyses of experience and language by postulating fictional characters, but before deciding to admit them these benefits must be weighed against the ontological costs. It is generally supposed that these costs are high, for admitting fictional characters, it is thought, involves postulating a new category of extremely strange and unusual entities, thus taking on a large burden compared with the parsimony which could be maintained without them. These worries should be taken seriously but should not be left unexamined. Does admitting fictional characters require positing a strange new category of beings bloating an otherwise spare ontology? Do the costs in terms of parsimony outweigh the benefits gained in analyzing experience and discourse? To properly assess the ontological costs of admitting fictional characters we must step back to ask how we can make such ontological decisions in a principled and consistent manner rather than on the basis of vague fears or aesthetic preferences for what seems to be a sparser ontological landscape.
PIECEMEAL ONTOLOGY VERSUS CATEGORIAL ONTOLOGY
Ontology is a two-part venture. The first task is to lay out categories in which things might be claimed to exist, without commitment to whether or not such categories are occupied. The second task is that of assessing what there really is. Contemporary ontologists typically – although not always – focus exclusively on the second task and proceed on a one-by-one basis to argue for or against allowing certain kinds of things, be they numbers, universals, acts of consciousness, or fictional objects, into our ontology.
If we are to postulate fictional characters at all, it seems advisable to postulate them as entities that can satisfy or at least make sense of our most important beliefs and practices concerning them. Often theories of fiction are driven not by an independent sense of what is needed to understand talk and practices regarding fiction, but rather by a desire to show how fictional characters may find their place in a preconceived ontology of possible, nonexistent, or abstract objects – to demonstrate one more useful application of the ontology under discussion, or to provide catchy and familiar examples. Instead of starting from a ready-made ontology and seeing how we can fit fictional characters into it, I suggest that we begin by paying careful attention to our literary practices so that we can see what sorts of things would most closely correspond to them. I thus begin by discussing what sorts of entities our practices in reading and discussing works of fiction seem to commit us to, and I draw out the artifactual theory of fiction as a way of characterizing the sort of entity that seems best suited to do the job of fictional characters.
WHAT FICTIONAL CHARACTERS SEEM TO BE
Fictional objects as I discuss them here include such characters as Emma Woodhouse, Sherlock Holmes, Hamlet, and Tom Sawyer – characters who appear in works of literature and whose fortunes we follow in reading those works. In our everyday discussions of literature we treat fictional characters as created entities brought into existence at a certain time through the acts of an author.
Far more intensive work has been done on analyzing fictional discourse than fictional experience. As in the case of intentionality, most of this work has been done in pursuit of the idea that there really are no fictional objects to which we can refer and so has been driven by a desire to avoid reference to fictional objects at all costs. But again the issue properly should not be conceived as whether we can get away without referring to fictional characters, but rather as whether we can offer a better theory of language by occasionally admitting reference to fictional characters. I argue that we can.
Many problems in speaking of fictional characters parallel those for thinking of them. Sentences such as “all parties to the discussion are speaking of the same character” express the tacit assumptions behind all critical discourse about how to understand and interpret literary characters. It seems that attempts to understand such sentences without referring to fictional characters, whether by appealing to the senses involved, the context of reference, or some combination of these, would run into problems parallel to those described above. Yet giving up the idea that such sentences could be true in a robust sense (not just that they were alike in thinking of nothing at all) would be to give up a great deal.
But rather than rehashing those problems for the case of language, I wish to focus on the problem that brought discussion of fiction into analytic philosophy: How to analyze statements apparently referring to fictional characters.
In the intuitive view initially presented, fictional characters are higher-level dependent entities, indeed entities dependent in a variety of ways on a variety of entities. We can now outline precisely what those dependencies are and see that, as a result of those dependencies, fictional characters turn out to be abstract artifacts, a kind of entity often encountered yet little acknowledged. To make their ontological status clearer, I begin by using the prior work on dependence to make the intuitive version of the artifactual theory more precise and then turn to investigate how such dependent entities fit into a general modal metaphysics.
DEPENDENCIES OF FICTIONAL CHARACTERS
The immediate dependencies of a fictional character are first, on the creative acts of its author or authors, and second, on a literary work. Clearly the dependence of a fictional character on the intentional acts of its creator or creators is a rigid historical dependence. Its historical dependence on certain forms of intentionality signals it as an artifact, an object created by the purposeful activity of humans (or other intelligent beings). We are certainly no strangers to artifacts; on the contrary artifacts from computers to cutlery to couches are those entities that surround us most immediately in our everyday life. Yet little has been done to incorporate them into a philosophical ontology, perhaps in part because of fears that conditions for their identity would be too thorny, and in part because too little has been said about dependence on the thoughts and practices of human beings.
One fear that inclines many to reject fictional objects is that we are liable to get ourselves into trouble either by falling into contradiction or by trying to constrain such unruly entities into the confines of a well-behaved theory. The motivation for rejecting fictional objects on the basis of their supposed intractability goes back at least as far as Russell's claim that Meinong's nonexistent objects are “apt to infringe the law of contradiction” and has gained contemporary popularity from Quine's characterization of unactualized possible objects (and presumably other so-called nonexistents) as “a breeding ground for disorderly elements,” making them unsuitable for proper individuation and untractable in philosophical theories.
Those who do attempt to offer identity conditions for fictional objects generally do so by reducing fictional objects to ideal abstracta, so that their identity conditions are reduced to those of more familiar and tractable entities. But such attempts inevitably miss important aspects of our ordinary practices in counting characters as the same or different. This difficulty is not unique to fiction: Indeed, some of the stickiest problems of identity arise for other cultural artifacts such as statues and monuments, musical works, and literary works, which are not identifiable with either basic physical concreta or with ideal abstracta and which thus seem to demand identity conditions different from those for either.
By conceiving of fictional objects as abstract artifacts we can offer identity criteria for fictional objects both within and across literary works that not only are as clear and precise as those we have for ordinary objects but also correspond closely to our practices in treating fictional characters as the same or different.
Discussions of fiction typically begin with the question of whether or not we must postulate fictional objects, with the defender of fiction attempting to establish that we absolutely cannot do without them, and the opponent attempting to show how we can manage to avoid postulating them through paraphrasing our apparent discourse about them and reconceiving our apparent experience of them. I believe that this approach to fiction is misguided on two counts.
It is misguided, first, to address the question of whether we should postulate fictional objects without first understanding what sorts of things they would be. We cannot see the potential costs and advantages of bringing fictional entities into our ontology until we have a clear conception of what sorts of entities fictional characters would be and how they would compare with other entities we might bring into our ontology. Vague fears that fictional characters would be too disorderly, too strange, so that postulating them would be liable to get us into trouble often drive decisions to avoid fictional objects at all costs. But we can only address whether such fears are grounded on the basis of understanding what these entities would be. Thus I propose that in Part One we postpone the question of whether or not there are such things as fictional objects, and begin by considering an easier question: If we were to postulate fictional objects, what would they be? In answer to this question I begin to draw out the artifactual theory of fiction.
Email your librarian or administrator to recommend adding this to your organisation's collection.