In a classic essay on types of legal fictions and their functions, Alf Ross observes that the variety he calls the “creative legal fiction” is “no more than a peculiar technique for the analogical extension of legal rules.”¹ The peculiarity, as Ross explains, consists in the vehicle that leads to the extension—a vehicle that takes the form of an asserted equivalence or identity:

To say that a barbarian is a Roman citizen amounts to extending for foreigners the application of the procedural rules that have hitherto been confined to Roman citizens. To say that Bordeaux is in Middlesex amounts to saying that the rules … hitherto … confined to claims originating in England , are now … extended … [to] claims originating in other countries. (Ross 2000, p. 222)

In these examples, the fiction arises because of the role of what Ross usefully calls a “proposition” that is “only posed, not affirmed” (Ross 2000, p. 231; see also pp. 228–229). The proposition may seem to make an affirmative statement, but because it is only posed (or suggested), it remains tentative, hypothetical—and as I argue below, because it remains in this limbo of the incomplete, its ability to yield any consequential effects is radically limited in a way that does not apply to doctrines.² If the proposition were affirmed, it would be converted into an ordinary doctrine, with the same potential for further extension and application that all doctrines have in common. To extend the ambit of the procedural rules by saying that non-Romans are also subject to them, or to extend the ambit of the jurisdictional rules by saying that an English court can hear claims originating outside of England, would involve no fiction at all, but would merely articulate a doctrinal conclusion. Nor is any fiction involved when a court reaches these conclusions after contemplating various rationales for and against adopting them. We encounter Ross’s “peculiar technique” only when a court arrives at a legal conclusion by justifying it according to the kind of premise illustrated in his two examples, a premise that might be regarded as a factual statement, although Ross resists that view by speaking of posed propositions .

The analogical extension of rules is a pervasive feature of legal argument and analysis, usually achieved without the vehicle that Ross isolates. The process itself is too commonplace to have been treated by any commentator in this area as involving the use of a legal fiction, which is taken to pick out a more distinctive set of legal concepts and operations.³ The distinction between legal fictions and reasoned conclusions tracks the contrast between metaphor and simile, differentiating the statement that a barbarian simply “is a Roman citizen,” and therefore is subject to the same procedural rules, from the statement that the same rules apply because barbarians resemble Roman citizens—that is, that the parties should be treated the same way because they share the same features in light of the rules’ purposes.⁴ In literary theory, and in research on cognitive psychology , metaphors are seen as more productive than similes, because the former are less confined to a particular ground of similarity (Gentner et al. 2000, p. 200). At least implicitly, similes usually suggest a particular feature that the two terms share; thus the implications they propose also depend on that feature, whereas metaphors, lacking this implied restriction, exert a dynamic and almost unrestrainable force. As Ellen Spolsky observes, they “breed promiscuously in the brain, producing analogies among unconnected or incommensurable ideas” (Spolsky 2003, p. 161).

In the case of the legal fiction, the generative potential operates in precisely the opposite fashion. The fiction (the legal version of the metaphor) is constrained in a way that doctrines (the similes) are not. If a jurist were to rationalize the application of identical procedural rules by first identifying the similarities between barbarians and citizens, the result would be to leave open the possibility of further doctrinal changes based on parallel reasoning : those similarities might warrant the treatment of barbarians as citizens in other contexts, where the same considerations matter. If a jurist forgoes the comparison (and its attendant rationale) and simply asserts that barbarians are citizens, the result is to foreclose any opportunity for extension.⁵ By replacing the simile with a metaphor and presenting the relationship as an identity, the jurist eliminates the generative effects that the analogy would promote.

We may bring this view of legal fictions into clearer relief by considering the distinction between fact and fiction taken from research on the psychology of reading. Rather than defining fictions according to their content or their means of inciting speculation, I suggest that we can distinguish fact and fiction according to the differentiations that readers make when encountering material they associate with one or the other. This approach has the benefit of drawing on empirical research rather than giving a definition laid down by fiat and justified by the proponent’s own intuitions, and it allows us to locate particular examples along a spectrum, rather than simply characterizing them as inside or outside the category of fiction. A proposition may be more or less fictional—or, to use the terminology in the psychological research, more or less artificial—according to the way the reader uses it. Even those who disagree with the following account of legal fictions, then, might find that it nevertheless offers the tools for providing an alternative account.

In recent decades, research on how readers use the material they glean from texts has distinguished in particular between material that is used just for the purpose of understanding the text that furnishes it, and material that is “incorporated into a person’s existing knowledge of the world” (Potts et al. 1989, p. 304). One of the most influential contributions to this discussion, by George Potts and his colleagues, showed that “artificial information”—that is, information that readers regarded as having been created specifically for the text at hand—was tightly compartmentalized, so as not to interact with information accepted as generally true (“real information”). Readers combined real information with existing knowledge to yield further hypotheses, insights, and beliefs , while restricting artificial information to the domain in which it originated, allowing it to produce new information only through interaction with other material within that domain. Attending to the functional effects of this distinction, Potts notes that “compartmentalized information … is difficult to retrieve and use … in a context that is different from the one in which it was learned,” and that while “organizing information as a unit unto itself may hinder the application of that information to new situations,” this means of storage “may actually serve to facilitate the retrieval of that information in a memory task that requires one to access only that limited body of newly learned information” (Potts et al. 1989, pp. 305–306).

While there are many contexts outside of fiction that offer “artificial information,” numerous studies undertaken in the wake of Potts’s research have shown that readers apply precisely this distinction to material taken from literary narratives (Rapp et al. 2013; Busselle and Bilandzic 2008; Prentice et al. 1997; Gerrig and Prentice 1991). By engaging in “source monitoring,” readers cabin some details absorbed from their reading, and fiction constitutes one of the main sources that prompts this response (Johnson et al. 1993). This research on how readers manage the information they absorb has the potential to clarify some points of contention in longstanding philosophical and aesthetic debates about “truth in fiction,” as Stacie Friend has recently shown. Drawing on the work of Potts and other scholars in this area, Friend argues that instead of attempting to specify the “necessary and sufficient conditions for fictionality ,” by identifying the features of a text or statement that mark it as fictional, we should instead “look at the practices that underpin our interaction with various texts” (Friend 2008, pp. 166, 165, emphasis added). Here I take the same approach and use Potts’s concept of artificial information to consider how compartmentalization is related to the management of legal fictions and legal artifice more generally.

Commentators on “truth in fiction” distinguish between details that are only “true in the story” and details that are also “real world facts” (Weisberg and Goodstein 2009).⁶ This distinction explains why, even if readers consider Sherlock Holmes to be a fictional character, they might nevertheless believe that different kinds of tobacco have a distinctively colored ash (Doyle 2005 [1891], p. 126), and more generally that Holmes’s deductive methods are applicable to everyday life. Law allows for a similar, but perhaps more finely calibrated, set of distinctions. While empirical research on compartmentalized and incorporated information in the legal field might reward further study, copious examples of those phenomena are already conveniently available, in forms that serve as useful counterparts to the versions that figure in the cognitive psychology literature, if we look to legal practices of argument and citation. Just as information in a literary narrative might be applicable only within the narrative, whereas information in a textbook might be applicable more broadly within the same field of inquiry, in law some kinds of compartmentalized information are limited to a particular case, and other kinds are limited to the legal arena generally. Just as textbooks speak of “frictionless planes, spherical planets, [and] … perfectly rational agents” (Frigg 2010, p. 257), courts speak of “rational basis review,” “estoppels,” and “consideration”—concepts that are artificial, not because they were created for a particular piece of writing, but because they were created for a particular area of inquiry . In both contexts, the value of the concept depends not on whether it accurately corresponds to something outside the field of analysis, but on how well it explains phenomena within that field.⁷ (As Wolcher (2010, p. 297) nicely puts it, “Legal form exiles all that is ‘extraneous’ in reality.” The corollary is that legal form adopts new expedients to fashion its own reality.) Finally, some kinds of legal information may be incorporated into the reader’s knowledge about the world. Borrowing from the philosophers, we may say that some details are “true in the case,” some are “true in law,” and some are also true in the world outside of the law.⁸

For the most part, legal convention requires that facts may be applied only within the case that presents them. Once they have been proved and treated as part of the record, the facts may be combined with the other established facts in the case, to produce legal conclusions, but may not be transposed to other cases. They may even be false in other cases. For instance, a defendant might escape liability in one dispute because the plaintiff could not establish certain facts, but might find herself liable when sued by another plaintiff for a claim arising out of the same incident because the second plaintiff introduced evidence unavailable to the first one, thereby creating a factual record that contradicted the record in the previous case. To become applicable to other legal disputes, the facts usually must be freed from their case-bound moorings by a doctrine that expressly permits the move, either through the workings of res judicata or by virtue of having been subject to judicial notice.

Unlike facts, legal doctrines —even those articulated for the first time—are taken to be potentially transposable to new cases.⁹ A court might decline to adopt the doctrine on any number of grounds, but not on the ground that the doctrine was by its very nature incapable of applying to other disputes. (A controversial exception involves the rule, in some jurisdictions, that treats foreign law as a question of fact.¹⁰) Few doctrines apply generally without qualification; the varieties of direct and analogical extension—and the constraints on them—are manifold, and hence there are endless gradations along the spectrum between restriction to a single case and complete integration across legal fields. Some doctrines can be generalized only within the same area of law; others can be generalized across common-law domains but not to constitutional law; and most are subject to more intricate limitations relating to the doctrine’s context, rationale, or historical origins. The techniques for justifying the extension of legal rules have generated a vast literature, and no useful purpose would be served here by attempting to enumerate or arrange them according to their tendency to make doctrines more or less transposable. Suffice it to say that whereas Potts emphasizes the cognitive burdens that make it hard to apply compartmentalized information to new situations, and the comparative ease of retrieving that information in its original context , the routine demands of legal analysis reveal a further level of complexity in light of the many contingencies that give doctrinal information an ambiguous status, making it arguable that the information may be incorporated, but not necessarily permitting it. Arguments for broader or narrower doctrinal expansion tax the ingenuity of lawyers on a daily basis, and these endless possibilities suggest that legal analysis is in large part a question of manipulating information to make it more or less artificial. Just as few doctrines are broadly “true in law,” so also few are only “true in the case” that sponsors them.

Finally, just as literary narratives may supply readers with information they accept unreservedly as facts to be incorporated, legal decisions also offer information of this order. Facts that are not necessarily applicable to another case might nevertheless be accepted by readers as revealing information about the parties, or about human behavior, that can yield new insights when integrated with other beliefs. Perhaps readers also incorporate information they find in the analytical parts of legal decisions. It is unclear to what extent doctrinal information is restricted to the legal realm—and by what kinds of readers (e.g., lay or professional, parties or would-be litigants). Empirical researchers have studied readers’ use of material encountered in various media—such as textbooks, biographies, novels, movies, and ads—but have not included legal decisions as possible sources of information. Because courts explain themselves by referring not only to doctrines and prior decisions, but also to details outside of the legal realm, and because legal doctrines themselves may incorporate assumptions about non-legal phenomena, judgments offer an intriguing prospect for further research. It might turn out that readers are willing to incorporate material from some of the doctrinal and analytical aspects of judgments.¹¹ If so, this result could contribute to our understanding of the referential qualities of legal language .

Readers’ practices of compartmentalization and incorporation, with respect to literary narratives and legal decisions, show how we might consider both forms of writing to include elements that fall at the fictional end of the spectrum (or the “artificial” end, in Potts’s terminology) and elements that fall at the factual (or “real”) end. This distinction provides a handle for reconsidering various phenomena that are often characterized as legal fictions, including doctrines such as constructive possession and constructive eviction, presumptions such as that everyone is assumed to know the law, and deeming provisions such as that a federal statute is deemed to have carried, from the day of its passage, whatever meaning the highest court in the land ultimately assigns to it. At first glance, it might seem that all of these provisions are fictional in the same way that compartmentalized aspects of stories are fictional. That is, it might be said that even when the content of these provisions is such that a reader could extract information from them and apply it to non-legal phenomena, the reader’s source-monitoring would prevent any such application. Hence one might conclude that these should all be characterized as legal fictions because they are treated as fictional in the same fashion as material that fiction-readers take to be “true in the story” but lacking any generalizable truth.

Closer scrutiny, however, shows that the analogy cannot be sustained. Insofar as these propositions have carrying power beyond the case in which they arise, they resemble the artificial information of the textbook, rather than the more restrictively compartmentalized information of the literary narrative. One might reply that what is “true in law,” no matter how widely true in that domain, is still a legal fiction so long as it fails to correspond to our experience of the world, and that this is precisely what makes it a fiction.¹² What that answer helps to show, however, is that deeming provisions, presumptions, and constructive modes of liability are neither more nor less generalizable than ordinary legal doctrines of all sorts.¹³