The aim of this paper is to lay the philosophical foundations for a computational theory of legal reasoning with and about fictions. Jurisprudential theories of legal fictions typically focus on the first question, and the way in which fictions are created by courts (and more rarely, legislators) to justify a certain legal outcome. The term “legal fiction” in this tradition is a meta-concept that is used to label and classify certain argumentative tropes, it is however not a term that is found directly in the reasoning of the courts or the texts of statutes. In contrast, we argue that this discussion can benefit also from an analysis of the way in which law reasons explicitly about the nature of fictional entities—a type of argument we find in particular in copyright law . In these legal texts, the term “fiction” is used directly by the courts, as part of the legal object language. Conversely, attempts to give philosophically sound and formally rigorous accounts of certain aspects of copyright law can benefit from the jurisprudential discussion of legal fictions. We aim to show that in particular, the recent interaction between Anglo-American analytical philosophy and the Austrian school of ontology provides a promising theoretical framework to understand the problems that both types of legal discourse about fictions have faced. It also suggests some answers, which might eventually lead to a computational and formal theory of reasoning with and about fictions in law. While this is the ultimate aim of our research, the discussion in this paper remains informal and tries to establish, in system theoretical terms, how the law thinks about fiction. We do this by explicating and analyzing some of the underlying intuitions that inform legal reasoning about fictions in copyright law, and indicate the structural similarities to jurisprudential questions about legal fictions.

In intellectual property law, the law address what it means to be a story or a fictional object within a story, what it means to “create” a fiction, what constitutes the identity of a fictional object and, most importantly for philosophers, under which conditions two fictional objects can be said to be the same. Following Quine’s famous dictum that there can be no entity without identity (Quine 1969, p. 23) , when discussing fictional objects we must first ask questions such as:

These philosophical questions surface in different ways but in the same form in copyright law. In a US copyright infringement case, Walt Disney Productions v Air Pirates,¹ the alleged infringement was the depiction of various Disney characters, including Mickey Mouse and Donald Duck, in the defendant’s adult counter-culture comic books. The defendant’s characters had the same names but were a “a rather bawdy depiction of the Disney characters as active members of a free thinking, promiscuous, drug ingesting counterculture”.² Were they, or were they not, in broad terms, the same as the Disney characters, and what does “the same” mean in this case? In an infringement dispute in Russia between the “Harry Potter” creator JK Rowling and the Russian writer Dmitri Yemets, author of the highly popular “Tanya Grotter” books threats of litigation by Rowling were not followed up with successful legal action. While Yemets’ argument that “The characters and the stories in the book are Russian folklore based on Russian culture and traditions” was therefore not tested in court, they show the type of argument that seems to carry weight when deciding if, while inspired by Potter the differences between the characters and the stories might be so substantive that they are not any longer “about the same” fictitious person, Harry Potter. In contrast, the Dutch courts have ruled that, despite all the changes, Yemets’ books did not create a new and sufficiently independent work of art (see Karjala 2006).³ In the language of the court, this was due to the similarities between the storylines. For us, this raises the additional question of the relation between “storyline” and “character”—can we infer for instance that if two storylines involving fictional characters are the same or nearly the same, that also the fictional characters depicted in them are the same or nearly the same? The final example is a dispute over the extent to which the “Sherlock Holmes” and “Dr Watson” characters could still be entitled to copyright protection. In December 2013 the United States District Court for the Northern District of Illinois decided that copyright in these characters had expired, but only for the characters as depicted in the older Conan Doyle novels. All aspects of their characters that were only mentioned in the later novels, set after Sherlock Holmes’ retirement, remain however protected.⁴ But how do we distinguish these two Holmeses? And if an author now wants to write a new Holmes novel, but is prohibited from mentioning almost everything pertaining to Professor Moriarty (who only rose to prominence in the later work Valley of Fear), how can we say that he is still writing about “the same” Holmes, given how much his character was formed through the interaction with his nemesis? Does this not render any new Holmes necessarily “incomplete”, that is lacking character traits and memories Holmes is “known to” possess, according to the canonical work? We will later see how various forms of “incompleteness” or ontological underdeterminacy are indeed diagnostic features that set fictional objects, both in copyright law and the theory of legal fictions, apart from ordinary objects.

Copyright law also helps us better to understand the difference between fictional and non-fictional objects. Assume that, living an isolated life in a Bavarian mountain hut, I had never heard about Harry Potter. There I wrote a book about a schoolboy, Heinrich Toepfer, whose parents were killed by a nameless, powerful dark wizard. He joins a school for wizards, makes friends with Ronald Wiesel and Hermione Amtmann and eventually overcomes the foe who killed his parents. In this case, I would not have violated any copyright belonging to JK Rowling, regardless of how similar the stories or characters are. Absent a causal link of copying from her creation to mine, no amount of identical characteristics between Harry Potter and Heinrich Toepfer make them “the same” person for the purposes of copyright infringement. The stories are “about” different fictional people. Physical objects are different. Assume that I had instead discovered the properties of a field that exists throughout space and breaks certain symmetry laws of the electroweak interaction. In this case, Peter Higgs and I would have described, independently of each other, the same field. Had our respective theories differed regarding the attributes they ascribed to that particle, one of us would have been wrong and not the discoverer of yet another different field that has just these properties.

As we have seen, questions of the identity of fictional characters feature prominently in both philosophy and copyright law. They are, however, not normally raised in the jurisprudential debate on legal fictions. We argue that, just as the philosophy of fiction takes the question of identity between fictional objects as its starting point, so too should the discussion on legal fictions. Can different jurisdictions recognise “the same” legal fiction, even when they have discovered/invented it independently? If, as some theories of fictional objects argue, the creator of a fictional entity has a unique role to play in determining its features (see e.g., Voltolini 2006; Thomasson 2003), how does this fit into the legal reality where one court may invent a legal fiction, and other courts are not just permitted but often requested to “fill in” its attributes and develop the concept over time? How many changes can we accept until we have to conclude that we are now referring to a “different” entity? Before we try to bring together the debate in analytical philosophy and copyright law, we will briefly consider the discussion on legal fictions. We identify a systemic problem in this discussion—the vanishing of its subject matter—and indicate how this problem in particular can benefit from linking together philosophical and legal doctrinal discussions on the nature of fiction more generally.

In his critical analysis of Lon Fuller’s seminal work on legal fictions, Kenneth Campbell argues that Fuller’s work, while particularly rich in examples, falls well short of stating a theory of legal fiction (Campbell 1983) . In particular, it fails to establish criteria that are necessary and sufficient to distinguish legal fictions from a whole variety of other phenomena, ranging from perjury to mistakes to mere metaphorical expressions. Fuller himself accepts this difficulty when he tries to analyse the relation of legal fictions to metaphorical word use on the one hand and presumptions on the other. Ultimately, he acknowledges that fictions may turn out to be a merely stylistic, surface phenomenon.

In a particularly interesting analysis, Frederick Schauer has recently reached a similar conclusion (Schauer in this volume, Chap. 6). He openly states what Campbell only hints at—that if we follow Fuller’s account, there might not be such a thing as a discreet phenomenon of “legal fiction” left. Schauer considers this outcome unsatisfactory, taking as a given the existence of legal ­fictions. He concludes, convincingly, that the problem of legal fictions cannot be ­separated from that of legal truth, and ultimately of truth simpliciter (Chap. 6). In ­Schauer’s words, legal fictions are therefore always “parasitic” on the gap between ­legal language and the legal ideal. Law has to use a language tethered in everyday ­language. If this were not so, “legal fictions” would simply be technical terms, “true in law”, but not corresponding to anything outside it. But law also has to use rules, and rules by their very nature cannot respond to the more nuanced ­differences that we find in the world they try to regulate. Fictions are the ­necessary glue that permits judges to fill the gap between a legal ideal and the inevitable limitations that the constraints on technical legal language impose. This leads to a normative account of legal fiction as a necessary part of “good interpretative practice”.

While there are sound intuitions behind this analysis, a comparative approach shows how precarious this solution is. In particular, the jurisprudential discussion of fiction in Germany demonstrates that the concept cannot possibly be dependent upon the act of juridical interpretation , since the very same issues also arise in a discussion of rational law making. In 1859, Gustav Demelius’ influential work on legal fiction demonstrated its importance for codified legal systems where the legislator and not the judge, avails themselves of this tool. In 1969 and 1976 respectively, Josef Esser and Dieter Meurer published extended monographs that explicitly put the discussion of legal fictions into the context of a theory of good law making (Esser 1969; Meurer 1976). Esser in particular developed an account of legal fiction that follows in many respects that of Schauer. For Esser, there cannot possibly be “true” or “epistemic” legal fictions; any attempt to claim that legal fictions involve an element of deception or error is for him a throwback to discredited nineteenth century jurisprudence that sees law as a set of abstract statements about causal relationships in a “legal world”. He writes:

The often-heard opinion that there are in addition to the “referring”, non-essential or formal legal fictions also “essential” or “real” legal fiction (which, by “obscuring legal truth”, force the judge to distort facts), has to be ascribed to the realm of pure fantasy. It can only be understood as the total misunderstanding of the normative character of law. To be “misled” presupposes that a descriptive judgement was intended in the first place. The proponent of this mistaken idea has therefore to assume that law is a system of descriptive statements (with truth claims). (Esser 1969, p. 28, translation by one of the authors)

Esser thus echoes Schauer’s notion that legal fictions are essentially normative and not descriptive. However, contrary to Schauer, they are not the result of ordinary language use in law, nor are they conceptually dependent on the judicial interpretive practice. This makes his analysis a more explicit statement of an idea that Fuller also ponders but does not pursue to its logical conclusion: that legal fictions might be a mere stylistic quirk. Fuller discusses the legal expression “X is deemed (in law) to be Y” and concludes that it functions in close analogy to a legal fiction properly so-called. However, the openness with which the language of the law here acknowledges that X is “only deemed” to be Y means that a necessary characteristic of a legal fiction, its falsity, is lacking. “X is deemed to be Y” cannot be false for a very similar reason why Esser’s “referring fictions” cannot be false —they do not even try to make a statement about the external world. Fuller concludes that “legal fictions” as developed in common law countries and “deemed to be” constructions in Roman law seem indeed to be functionally identical. This leads to the conclusion that any difference in surface expression is an issue of legal style and not ontological substance.

At this point, we can identify a common problem in jurisprudential theories about legal fictions. The very moment we try to take legal fictions seriously and focus our analytical gaze upon them, they seem to disappear into nothing. Rather than describe them as “parasitic” as Schauer does, a better metaphor might be that of an optical illusion, such as the Troxler effect, where we have difficulty focusing our gaze on a “floater” in our field of vision. For Bentham, legal fictions are very much the villain of the piece, and are “to justice exactly as swindling is to trade” (Bentham 1938, p. 235). If they are villains, though, then they are very much like the villainous Weeping Angels in the universe of Dr Who: terrifying, fast, dynamic and ultimately lethal, but only when not looked at. The moment we focus our eyes on them they become just an immobile part of the decoration, rendered harmless and ineffectual by the very process of observing them.

We can get an initial idea why this might be the case by looking at some proposed diagnostic criteria for legal fictions:

The first three criteria all use a privative definition; they tell us more what legal fictions are not than what they are. Legal fictions are not true, they are not lies and they are not mere errors. Following Fuller’s analysis, Schauer concludes that, because the concept of legal fiction is parasitic upon the concept of legal truth, we cannot determine what legal fictions actually are unless we have a theory of legal truth. We argue that the intuition behind this idea is sound, but that the way it is expressed conflates two crucially different issues that the separate philosophical debate on fictions can help us differentiate. First, “truth” and “falsity” are properties of propositions, not of objects. The literature on legal fictions, including Bentham, Fuller and Schauer, tends to conceive of legal fictions as statements or doctrines. This leads naturally to thinking of them as something that is “not true”. But this is a problematic starting point for whatever one thinks about legal fictions , it seems obvious that there are many statements using legal fictions that are literally true. “Legal personhood of companies is a legal fiction” is an obvious candidate. “Legal persons do not have a right to vote in elections in the US” is a more interesting, intuitively true statement. Since we can always make true statements about objects that, in one way or another, do not “exist” in reality, it is not therefore the true/false dichotomy that should concern us, but a question of ontology: to what, if any, part of reality do terms for legal fictions (such as “legal person”) refer? A related question of metaphysics now arises: what properties do these objects need to have, if they exist, to do what they do? Legal fictions, being objects, cannot be lies, mistakes or true accounts of the law. They are of a different syntactic category altogether. Rather, the question is what terms for fictional objects contribute to the truth value of the sentences in which they appear and whether this differs systematically from the contribution other objects make.

Philosophical discussions on the nature of fictions tackle this issue head on. Bertrand Russell’s highly influential work “On denoting” (1905) is a case in point: in his example sentence “The King of France is bald”, the term “The King of France” behaves differently from “The Queen” in “The Queen of England is bald”. Both sentences are false, but only in the second case can we infer that the monarch has hair. Russell proposes “unpacking” the fictional “King of France” as a complex statement that eliminates the need to talk about fictional entities, rendering it as: “There is one person only who is king of France and bald”. Although Russell’s eliminative approach remains influential in the philosophical discussion of fiction, it faces the problem that his eliminative stance means that sentences that we intuitively judge to be true turn out to be false or meaningless.

While his analysis resonates with accounts of fictions that relate them to lies or mistakes, transposing it to legal fictions faces problems and creates difficulties. Russell’s analysis tries to understand fictional objects as the opposite of physical objects. However, it is not only legal fictions, but also many other pieces of our legal “deontic furniture” (Hage 2012) that are different from physical objects. Physically, money is just a piece of coloured paper, what makes it money is that it “counts as” a way of making payments. Registering a land purchase does not alter its physical properties, but henceforward it “counts as” property. Searle’s “counts as” conditionals are central for the creation of the “legal world” and its institutions (see e.g. Boella and van der Torre 2006). “Counts as” fulfils the same linguistic function as the “deemed to be” construction for legal fictions discussed above. If we were to eliminate legal fictions Russell-style, we risk eliminating most of the law’s institutions along with them. The implicit physicalism that underpins his analysis is unsuitable for an analysis of legal ontology.

A possible way out for the eliminist is to focus on the act by which fictions are created. For Searle, “counts as”, like its cousin “deemed to be”, are speech acts. We could consider the performative act by which judges create legal fictions as an “unauthorised” attempt to add objects to our ontology. Historically, this is the line that Bentham took, and it continues to influence the common law analysis of legal fiction. But in Searle’s analysis, performative acts without the right authority simply fail: the judges in such a case may have tried to create a new object, but failed to do so. In such a case, “legal fictions” are simply errors. Here, too, legal fictions ultimately “disappear”: there can be no such thing. Nor does this analysis account for the creation of legal fictions by the legislature, which by definition has all the authority needed. However, as our comparative analysis has shown, legal fictions can be created through statute.

By blurring the distinction between mistakes and legal fictions, we also fail to account for the understanding of fiction within the legal community. For example, lawyers understand intuitively that there is a difference between someone claiming that legal persons in Scots law must have only female partners (a mistake) and a judge granting a company the right to file charges against a debtor (a “correct” use of the legal fiction that employs the concept). If “legal person” had no referent, both sentences would be equally wrong.

Finally, Russellian eliminativism also makes it impossible to distinguish different legal fictions from each other. If I fail to build a house, and also fail to plant a tree, then I do not create two new and distinct objects—“my failed-to-be built house” and my “failed-to-be-planted tree”—that have properties distinguishing them from each other. In both cases, I created the same thing, that is a no-thing (for further discussion, see Donnellan 1974). From a Russellian point of view, the house I did not build is the same “thing” as the tree I did not plant. Similarly, all legal fictions would refer to “the same” thing, the empty set. Ejectment, legal personhood and becoming steward of the Chiltern Hundreds all become “the same” thing—that is nothing. A theory of legal fictions, or a history of them, becomes impossible.

We therefore face a dilemma: because our ontology of legal fictions is parasitic on our legal ontology simpliciter, any attempt to determine legal fictions is in danger of collapsing them into either “normal” legal objects; or they are seen as cognitive mistakes or lies, and with that no objects at all. Yet neither route is appropriate to explain the semantic role legal fictions actually play when used in legal reasoning.

To find a way out of this dilemma, we need to circumvent the ontological question at least temporarily, and find some data that is independent from a discussion of general legal ontology. The discussion of literary fictions in copyright law and philosophy hopes to achieve just that. The first thing we can learn from it is to separate the “ontological” question of whether fictions exist from a discussion of their “metaphysical nature”—that is, an attempt to determine what characteristics they have to have if they exist (see Thomasson 1999). The way to determine their nature in turn is to analyse what contribution they make to the sentences in which they appear. A systematic analysis of copyright decisions would therefore be highly desirable. We cannot offer such a detailed analysis in this paper, however, as it would require a discussion of the underlying legal framework, something that must be left to future work. Instead, we will discuss the three most promising philosophical theories on fictional objects and will highlight, through a few examples, where the connection between copyright law, analytical philosophy of fiction and jurisprudential discussions of legal fictions are the most interesting or promising.