The Identity of Ficta and the Place of Authors

I. Introduction

In this article, I present a dilemma concerning the relationships between the identity conditions of fictional entities (or ficta, in short) and their authors. I argue that both horns of the dilemma give unpalatable results. If one accepts that such relationships may legitimately figure in the identity conditions of ficta, three problems arise: the Problems of Necessary Authors, Circularity, and Redundancy. If one denies this possibility, one must reject the intuition that ficta depend on their authors for their own identity. Moreover, one cannot exploit relationships with authors to fix the distinctness or identity of indiscernible and inconsistent ficta.

Finally, I show that it is plausible to endorse one horn of the dilemma without accepting the troublesome assumption responsible for its unpalatable results – namely, that if authors fix, or contribute to fixing, the identity conditions of ficta then certain relationships with authors should partake in those very conditions. This move clarifies how and to what extent ficta depend upon their authors and solves the problems connected with both horns. However, to account for indiscernible yet distinct ficta, it is necessary to introduce primitively individuated and individuating entities – namely, ‘traces’.

Section II sets the stage. Section III presents the dilemma and its horns. Section IV motivates my solution.

II. Setting the Stage

Fictional stories are collections of propositions endowed with narrative unity and put together by their authors without the intent to deceive. For example, Stefan Zweig’s The Royal Game is a short novel written before 1942. It is a fictional story about two fictional chess players, Mirko Czentovic and Dr. B. It has narrative unity: it focuses on their lives, their encounter, and the match they play during a cruise from New York to Buenos Aires, all narrated by an anonymous passenger. When writing the novel, Zweig did not aim at convincing readers that the match actually took place, or that Mirko Czentovic and Dr. B existed. Zweig was inspired by real events of the first half of the twentieth century, such as the Nazi occupation of Austria. But his intent was to narrate a story no one was expected to believe true. Presumably, by narrating the story, Zweig aimed at exploring the deep meaning of chess and the tragedies of his time and of his own life.

The characters in a fictional story are all those entities to which some features (that is, properties or relations) are ascribed according to the story. Mirko Czentovic, Dr. B, and the anonymous narrator are characters in The Royal Game. But even the Austrian Chancellor Schuschnigg is a character, since features are ascribed to him in the story.

A fictional character in a fictional story (or a fictum) is a character that does not exist independently of any story. Mirko Czentovic, Dr. B, and possibly the anonymous narrator are ficta. Schuschnigg is not – or so it seems.¹

By using ‘entities’ when talking of ficta, I aim at remaining as neutral as possible. Ficta may be taken as bona fide mind-independent or mind-dependent existing objects, as non-existing ones, as properties or conjunctions of properties, and so on.²

Moreover, I assume that there are ficta. However, one may also adopt an anti-realist attitude, holding that there are ficta only according to fictional stories. From this standpoint, ficta should be treated as, or replaced with, fictional names³ or with (portions of) the contents of pretences or make-believe games, that is, with ‘surrogates’ of ficta.⁴ A ‘surrogate’ of a fictum is whatever replaces that fictum within the anti-realist perspective and plays at least some of the roles usually attributed to it.

Everything I shall claim here may be easily translated in such anti-realist terms, by talking about ‘surrogates’ or by introducing them to fix claims of identity or distinctness apparently about ficta. For the sake of simplicity, however, I shall assume here that there are ficta and that they are objects.

By using ‘features’ to refer to the properties and relations ascribed to characters according to fictional stories, I do not endorse any specific theory of properties and relations, or any realist attitude towards them.

The original author of a fictum is the entity responsible for the fictional story or stories in which the fictum first appeared. Zweig is the original author of both Mirko Czentovic and Dr. B, which, to my knowledge, did not appear in any earlier story. In some cases, however, it is very difficult – if not impossible – to single out the original author of a fictum, and such an author may even be a group of individuals.

The original author of a fictum is supposed to perform some relevant acts in order to be its original author. Such acts concern the relevant fictional story or stories and, presumably, the fictum itself. For example, in order to be the author of Dr. B, Zweig must perform certain acts, such as selecting the relevant characterizing features of Dr. B,⁵ creating Dr. B, introducing the first use of the fictional name ‘Dr. B’, and so on – depending on one’s theory of ficta.

The non-original authors of a fictum are those responsible for fictional stories in which the fictum seemingly appears, though they are not responsible for the fictional story or stories in which the fictum first appeared. From now onwards, by using ‘authors’, I shall refer to both original and non-original authors.

When ficta figure in different stories (possibly by different authors) or in different portions of the same story, it is useful to ask whether they are identical or distinct. More precisely, suppose that someone (call her ‘Jane Horwich’) writes a sequel to The Royal Game, in which Dr. B meets Mirko Czentovic again, but now Dr. B has a slightly different past from the one Zweig ascribed to him. In this case, it is natural to ask if Dr. B in The Royal Game is the same fictum as Dr. B in Horwich’s sequel. Answering this question requires investigating the identity conditions of Dr. B.

With respect to the identity conditions of a fictum, the following schema is usually adopted:

The relevant necessity is metaphysical necessity, that is, necessity based upon the nature of ficta themselves or of their ‘surrogates’ (for instance, fictional proper names or portions of contents of pretence or make-believe games). P should be replaced by a necessary and sufficient condition for the identity of x and y, or a set of jointly necessary and sufficient conditions, and it should not quantify over ficta; otherwise, it would presuppose the very identity conditions it is meant to clarify.

Some authors only accept weaker schemas for the identity conditions of ficta, which express merely sufficient or merely necessary conditions.⁶ I shall assume, however, that, if possible, it is preferable to adopt (INS).

Other desiderata include that, in specifying P, one should avoid regresses, circularity, brute facts of identity or distinctness, potential counterexamples,⁷ and the positing of ad hoc and controversial entities. One should be as informative as possible and able to account for most, if not all, troublesome cases of identity and distinctness among ficta.

Let me now assume that A stands for some relevant act, intention, relation, or feature involving the author of a fictum and that may figure in P. Namely, A stands for an author-involving feature (from now onwards, an AI-feature) that may figure in the identity conditions of ficta.

For example, A may stand for the following: Horwich’s using ‘Dr. B’ with the (conscious) intention of identifying the referent of that name in her own story with the referent of ‘Dr. B’ in Zweig’s The Royal Game. Or, with Zweig’s own story, A may stand for Zweig’s using ‘Dr. B’ in a certain portion of The Royal Game with the (conscious) intention of identifying its referent with that of ‘Dr. B’ in another portion of that story. A may also stand for Horwich’s intention of importing Zweig’s Dr. B in her own story.⁸

In all of these cases, it is necessary to better specify A without already assuming that ficta x and y are identical. For example, Horwich’s ascribing to her own Dr. B a relevant set of features ascribed to him by Zweig in The Royal Game should be replaced with the following: Horwich’s ascribing to at least and at most one fictum in her own story a relevant set of features ascribed to at least and at most one fictum by Zweig in The Royal Game.

Of course, not all AI-features are equal when it comes to figuring in the identity conditions of ficta. Presumably, those involving the original authors matter more.

Two final caveats. First, I shall not be concerned here with the identity or distinctness between characters in fictional stories and real entities. Second, one may suggest relaxing (INS) by replacing numerical identity between ficta with continuity or qualitative sameness between possibly distinct ficta, that is, with (possibly distinct) ficta being continuous along some common (spatio)temporal or causal pattern, or with (possibly distinct) ficta being ascribed all and only the same features. However, even when one appeals to continuity or qualitative sameness, one is not dispensed with the additional – and more fundamental – task of determining whether such ficta are actually identical or distinct. Both continuity and qualitative sameness are compatible with numerical identity as well as distinctness. Moreover, when defining the conditions for continuity or qualitative sameness, one must ultimately provide adequate conditions for the numerical identity or distinctness of those very ficta and/or of their ‘surrogates’.

III. The Dilemma

Recall:

The dilemma arises when we ask whether AI-features can legitimately figure in P.

Suppose that AI-features cannot legitimately figure in P, that is, in the identity conditions of ficta. This conflicts with a central intuition underlying many theories of ficta – namely, that ficta (or their ‘surrogates’) depend on their authors in crucial respects (at least, on their original authors). For example, Dr. B would not have existed without Zweig’s (or some other author’s⁹) intention of introducing this character in The Royal Game. If ficta (or their ‘surrogates’) depend on their authors in such respects, then authors fix, or contribute to fixing, their identity conditions through AI-features. Thus, Zweig (or some other author) – or, better, certain AI-features involving them – fix, or contribute to fixing, the identity conditions of Dr. B.¹⁰

Second, AI-features are often sufficient and/or necessary to fix the identity or distinctness of ficta in certain circumstances, or are part of sufficient and/or necessary conditions for doing so. Thus, AI-features should figure in the identity conditions of ficta.

Here are some interesting cases. Consider two fictional twins that share all and only the same features in a fictional story – maybe because they have few features according to that story.¹¹ Or consider the fictional members of a fictional crowd, characterized only as distinct members of that crowd.¹² In such cases, the features these ficta have according to the relevant stories are not sufficient to fix their distinctness. What is required, in addition, is an AI-feature, such as the author’s intention to create (or select, but I shall omit selection in what follows) numerically distinct though indiscernible ficta. For example, in the case of fictional twins, the following AI-features may be in place: the intention to create one fictional twin and the intention to create another fictional twin. These intentions are strictly connected. Alternatively, if only one intention is present (namely, the intention to create two fictional twins), one may distinguish different ‘parts’ of that intention or of its content: one leading to the creation of the first twin and one to the creation of the second twin. These intentions, or their relevant ‘parts’, may account for the distinctness of the fictional twins.

Additionally, consider a philosophical variant of Jorge Luis Borges’s Pierre Ménard.¹³ Ménard is an author who writes a story about a fictum that is qualitatively indiscernible from Miguel de Cervantes’s Don Quixote, without knowing anything about the latter. Ménard’s Don Quixote and Cervantes’s Don Quixote are thus indiscernible with respect to the features they have according to the relevant stories. If one wishes to hold that they are distinct ficta, one must appeal to distinct relations with distinct original authors. In this case, the following AI-features should figure in their identity conditions: being created by Ménard and being created by Cervantes.

Finally, a single fictum may have inconsistent features across different stories, or even within the same story. For example, Dr. B may lack the feature of being a politician according to Zweig’s The Royal Game, yet possess it in Horwich’s (fictional) sequel. In this case, only Horwich’s intention of importing Zweig’s Dr. B in her story can account for their identity. By contrast, if one considered only the features possessed by the ficta in the respective stories, one would conclude that there are two distinct ficta, since they are discernible.¹⁴

Thus, one horn of the dilemma – that AI-features cannot legitimately figure in the identity conditions of ficta – should be avoided. Turn now to the other horn – namely, that AI-features can legitimately figure in those conditions. There is an easy argument in favour of this option: as we have seen, AI-features fix, or contribute to fixing, the identity or distinctness of ficta.

Moreover, the following assumption seems acceptable:

The obvious conclusion is that AI-features can legitimately figure in the identity conditions of ficta. As I shall argue in Section IV, however, (AI) is questionable. For now, let me present three problems that arise from this horn.

The first problem is the Problem of Necessary Authors. Suppose that Ménard’s relations with his Don Quixote figure in its identity conditions. This seemingly entails that that fictum cannot but have Ménard as its author. If it had been authored by someone else, it would not have been that very fictum – Ménard’s Don Quixote – but someone else’s. Thus, Ménard’s Don Quixote is necessarily tied to Ménard. More generally, if AI-features involving specific authors figure in the identity conditions of ficta, ficta seem necessarily tied to their authors: they cannot but have them.

One could swallow this consequence – controversial as it is. Alternatively, one might resist it by suggesting that the relevant AI-features do not involve specific authors but rather some author or other. For example, in the case of Ménard’s Don Quixote, the relevant AI-features would not involve Ménard but merely some author or other.¹⁵ Yet such AI-features would not suffice to distinguish Ménard’s Don Quixote from Cervantes’s Don Quixote, since both would involve the same AI-features of being authored by some author or other.

To be clear: in this context, the Problem of Necessary Authors only amounts to claiming that necessary authorship is philosophically controversial and requires additional arguments. Ceteris paribus, then, it is preferable to avoid being committed to necessary authorship just by including AI-features in the identity conditions of ficta.

The second problem for this horn is the Problem of Circularity. As we have seen, identity conditions for ficta should avoid circularity, that is, they should avoid assuming or presupposing in P the identity or distinctness of the ficta at stake. Yet many AI-features presuppose, in their own identity conditions, the identity of the ficta they are meant to ‘ground’. For example, Zweig’s intentions concerning Dr. B (or his ‘surrogates’) are mental acts with a certain object – namely, Dr. B. These acts are the acts they are also because they concern that very object and no other. Thus, Zweig’s intentions concerning Dr. B (or his ‘surrogates’) seemingly include Dr. B (that very object) in their own identity conditions. If such intentions were then invoked in the identity conditions of Dr. B (that is, in P), a problematic circularity would arise.

One may avoid the Problem of Circularity by looking at AI-features that do not include their own objects in their own identity conditions. Yet it is difficult to find such AI-features. Suppose that we adopt an anti-realist stance towards ficta. There remains the problem of relating the author of a fictum with the ‘right surrogate’ of that fictum (for example, the ‘right’ fictional name, such as ‘Dr. B’). The ‘surrogate’ is the ‘right’ one only because it is related – or it seems to be related – with the ‘right’ object of thought or portion of mental content in the author’s mind. Thus, the ‘right surrogate’ must include some specific content or object in its own identity conditions. But, from the anti-realist standpoint, that very content or object was meant to be replaced with the ‘right surrogate’. This strategy therefore fails. And if it fails for anti-realist ‘surrogates’ it is unlikely to succeed for ficta on a realist view.

Alternatively, one may appeal to generic AI-features, for example Zweig’s using ‘Dr. B’ with the (conscious) intention of referring to some fictum or other with certain features. But this is insufficient to distinguish between qualitatively indiscernible ficta such as Pierre Ménard’s and Cervantes’s Don Quixotes, since both partake in the same generic AI-feature.

Finally, one may invoke singular AI-features, such as Zweig’s using ‘Dr. B’ with the (conscious) intention of referring to at least and at most one fictum with certain features. The problem is that Zweig cannot guarantee by himself that this condition is satisfied and that Dr. B actually has this feature. Whether the ‘at most’ clause holds does not depend solely on Zweig. Zweig may well do everything within his own power to ensure that only one fictum called ‘Dr. B’ appears in his own works, but another author may, unbeknownst to Zweig, intend to use ‘Dr. B’ for at least and at most one fictum with the very same features. In this case, either Zweig and the other author independently create the same fictum – which is problematic – or there are two ficta, in which case the ‘at most’ clause is violated.

One may reply that the intentions at stake are distinct, because one involves Zweig and the other involves the other author. But this move then falls prey to the Problem of Necessary Authors.

Third, there is the Problem of Redundancy. AI-features seem redundant in comparison with their results or products, at least when it comes to partaking in the identity conditions of ficta. It is only those results that should figure in such conditions.

Indeed, suppose that the relevant AI-feature to figure in the identity conditions of Dr. B is Zweig’s using ‘Dr. B’ with the (conscious) intention of identifying a fictum in one portion of The Royal Game with a fictum in another portion. What matters in this case is that Zweig successfully uses the name with that intention, that is, that his act reaches its relevant result. This implies that the relevant result – the successful use – should itself figure in the identity conditions of Dr. B. If, for any reason, that result were not reached, Zweig’s act would not be relevant to figure in those conditions, since it would not successfully contribute to fixing the identity of Dr. B.

In general, the results of AI-features should already be legitimately included in the identity conditions of ficta, that is, in P. On the contrary, the AI-features leading to those results matter to P only insofar as their results are already and legitimately included in it. In this case, AI-features are redundant with respect to P: once their results are included (as they should be), it is already established that those results have been reached by the relevant AI-features, and there is no need to include the AI-features in addition to their results.

Alternatively, AI-features are non-fundamental components of the identity conditions of ficta, whereas their results are fundamental ones. Or, if identity conditions are meant to capture the essence of ficta, the results of AI-features may belong with the constitutive essence of ficta (namely, in what directly defines the essence of ficta), whereas AI-features themselves can figure only in the consequential essence of ficta.¹⁶

In sum, if AI-features are allowed to figure in the identity conditions of ficta, one incurs the Problems of Necessary Authors, Circularity, and Redundancy. If they are excluded, however, this move is at odds with a powerful intuition underlying many theories of ficta, according to which the identity or distinctness of ficta depends in some way on their authors. Moreover, difficult cases – such as fictional twins, members of fictional crowds, Ménard’s and Cervantes’s Don Quixotes, and ficta with inconsistent features – remain unaccounted for.

IV. Fixing without Partaking

My solution consists in accepting that authors, through specific AI-features, may fix or contribute to fixing the identity conditions of ficta. Thus, Zweig’s intentions also fix or contribute to fixing the identity conditions of Dr. B by making it the case that certain other features – such as being an amateur chess player, a prisoner of Nazis, being called ‘Dr. B’ – partake in those conditions. These features are jointly necessary and sufficient to reidentify Dr. B across different stories or different portions of the same story.

However, this does not imply that authors and AI-features figure in the identity conditions of ficta. In particular, I deny:

Consider the following cases. Suppose that part of my identity conditions is that I am a human being and that I am endowed with a certain genetic code. My parents obviously fix these identity conditions: they make it the case that I am human and that I have that genetic code. Thus, there is some Parent-Involving feature (PI-feature) – some relation with my parents – that fixes, or contributes to fixing, these conditions. Yet this does not imply that the PI-feature itself should figure in my identity conditions.

First, at least in principle, my being human and endowed with a given genetic code may have been due to other parents, and thus to other PI-features.¹⁷

But let us grant that it is necessarily the case that I have those features – being human and having a certain genetic code – only if I have the relevant PI-feature involving my own parents. Even so, the PI-feature at stake is redundant in comparison with its results – namely, the former features of being human and endowed with a certain genetic code. Thus, the PI-feature is at best a non-fundamental part of my identity conditions. More precisely, it is not part of my essence – of what most fundamentally defines my identity – but at most a proprium, a necessary accident of my essence, something that is necessitated by it without partaking in it.¹⁸ Alternatively, the PI-feature is only part of my consequential essence, not of my constitutive essence.¹⁹

Consider another case: Michelangelo’s David. This statue has certain features in its identity conditions – for example, representing young King David or having a certain shape. Moreover, Michelangelo is obviously responsible for these features being included in David’s identity conditions, since he sculpted David with certain intentions. There is thus a Sculptor-Involving-feature (SI-feature) – a sequence of Michelangelo’s intentional acts – that fixed or contributed to fixing David’s identity conditions. Yet this SI-feature need not itself be part of those conditions. Thus, Michelangelo need not figure in them. Another sculptor and/or another SI-feature may have been responsible for fixing David’s identity conditions – or at least this scenario is clearly conceivable and therefore prima facie metaphysically possible. It is up to those who deny this possibility to argue that, although conceivable, it is not metaphysically possible.

Even granting that David has those identity conditions only if Michelangelo’s SI-feature is in place, that SI-feature remains redundant relative to its results – namely, the features included in David’s identity conditions. It is only those results that must figure in David’s identity conditions – or, more precisely, in its essence. Michelangelo’s SI-feature may at most be a non-fundamental part of those identity conditions, or a proprium, or part of David’s consequential essence.

Mutatis mutandis, these thoughts apply to AI-features. That Zweig’s AI-features fixed, or contributed to fixing, Dr. B’s identity conditions does not entail that those very AI-features must figure in those conditions, or that they must fundamentally partake in them – that is, be part of Dr. B’s essence. Thus, (AI) is false. Or, if one holds that at least some AI-features may non-fundamentally figure in the identity conditions of ficta, then (AI) may be true but the following stronger version of (AI) is false:

A variant of (AI’) may be constructed in terms of constitutive and consequential essences. For simplicity, however, I shall deny (AI) simpliciter, assuming that whatever partakes in the identity conditions of a fictum does so in a fundamental fashion or as part of its constitutive essence.

To see the same situation from another standpoint, assume that Dr. B’s identity conditions are exhausted by a set of features, such as being an amateur chess player, a prisoner of Nazis, and being called ‘Dr. B’. It is reasonable to hold that, to know the essence of Dr. B, one must know the essence of such features. That is, to know what Dr. B is, or to provide a ‘real definition’ of Dr. B (as a fictum), one must also know, for example, what being an amateur chess player amounts to.

Now assume that Zweig’s intentions fix, or contribute to fixing, this set, for example, by making it the case that exactly those features – and no others – exhaust Dr. B’s identity conditions. To know the essence of such features, it is not necessary to know the essence of Zweig’s intentions, that is, their ‘real definitions’, including what being an intention amounts to (or a certain sort of intention or an intention with a given content or with a given ‘bearer’). One could know, for example, the essence of being an amateur chess player without knowing the essence of Zweig’s intentions and even without knowing anything about Zweig. Thus, Zweig’s intentions fix, or contribute to fixing, the relevant set of features without being involved in the essence, and hence the identity conditions, of those features, and without Zweig’s intentions partaking in that set. Fixing the identity conditions may amount to creating a fictum, for artifactualists, or may be entailed by creation.²⁰ Nothing changes if one adds that that very set of features cannot but have been fixed by those very intentions (and no others) and by Zweig’s intentions (rather than anyone else’s).

Through these examples, I do not wish to deny that some PI-, SI-, or AI-features may fundamentally partake in the identity conditions, and thus the essences, of organisms, statues, or ficta. Namely, I do not wish to deny that the former may be part of the essences of the latter. What I deny is that this must be so merely because such features fix, or contribute to fixing, their identity conditions. Yet, even if one holds that PI-, SI-, and AI-features are essential to organisms, statues, and ficta, one should still face the three problems presented in Section III: the Problems of Necessary Authors (and Parents, and Sculptors), Circularity, and Redundancy. The proposed solution is neither ad hoc nor unmotivated, since it applies equally well to entities other than ficta, such as organisms and statues (or artefacts more generally).

I now turn to showing how my solution addresses the Problems of Necessary Authors, Circularity, and Redundancy.

First, my solution is compatible with both contingent and necessary authors. AI-features involving specific authors (such as Zweig) need not be included in the identity conditions of ficta; ficta could therefore have been authored by different authors. At the same time, ficta may still be necessarily tied to their own authors. As noted above, I do not deny that some AI-features may (fundamentally) partake in the identity conditions, and thus the essences, of ficta. What I deny is that this must be so just because AI-features fix, or contribute to fixing, those conditions. Moreover, a fictum may be necessarily tied to its actual author and to the relevant AI-features without the latter being part of the essence of the former. Some AI-features may instead be necessary accidents (propria) of ficta.

The Problem of Circularity is avoided insofar as AI-features – which may include ficta in their own identity conditions – do not figure in the identity conditions of those ficta. However, a new version of this problem may resurface. One may concede that some AI-features (such as certain intentions of Zweig) include ficta in their own identity conditions, and also concede that those AI-features fix, or contribute to fixing, the identity conditions of the ficta at stake. This seems to imply, by transitivity, that ficta fix their own identity conditions.

In reply, one may hold that ficta and AI-features actually stand in two distinct and non-equivalent dependence relations. It is one thing (that is, one dependence relation) to figure in something’s identity conditions and it is another thing (that is, another dependence relation) to fix, or contribute to fixing, the identity conditions of something (for example, by bringing something into existence or by selecting the relevant features). Crucially, that something figures in the identity conditions of something else need not entail that it fixes, or contributes to fixing, the identity conditions of the latter. And that something fixes, or contributes to fixing, the identity conditions of something else need not entail that it figures in the identity conditions of the latter. Thus, ficta may figure in the identity conditions of certain AI-features, while those very AI-features may fix or contribute to fixing the identity conditions of ficta. This does not imply, by transitivity, that ficta also fix, or contribute to fixing, the identity conditions of AI-features and therefore – by transitivity – their own identity conditions.²¹ An alternative reply will be explored shortly, when dealing with indiscernible yet distinct ficta.

The Problem of Redundancy is avoided by maintaining that AI-features need not be part of the identity conditions of ficta together with their results. It is only the latter that must be included in those conditions. Thus, AI-features are not redundant with respect to their results.

My solution preserves the intuition that ficta (or their ‘surrogates’) depend on their authors in certain crucial respects. AI-features may fix, or contribute to fixing, the identity conditions of ficta without themselves partaking in them. Moreover, the solution is compatible with theories according to which the identity conditions of ficta involve no AI-feature at all. And my claims are compatible with AI-features playing no role in fixing those conditions. I have only suggested that AI-features can fix or contribute to fixing the identity conditions of ficta without partaking in them.

Moreover, AI-features may be relevant for ficta only when it comes to something different from their identity conditions – for example, when a certain fictum (with given features) is ‘selected’ by its author to partake in a story, even if that fictum is a (non-existent) object characterized by those features, and possessing its identity conditions, independently of the author.

With distinct yet indiscernible ficta, however, my solution faces a difficulty. Recall that distinct yet indiscernible ficta are ficta such as fictional twins, fictional members of fictional crowds, and Ménard’s and Cervantes’s Don Quixotes. In Section III, I suggested that their distinctness may be fixed by distinct AI-features figuring in their identity conditions. For example, Ménard’s Don Quixote is distinct from Cervantes’s Don Quixote because two distinct AI-features partake in their identity conditions – namely, being created by Ménard and being created by Cervantes. Mutatis mutandis, something similar may happen in the other cases, either by appealing to distinct AI-features or to distinct ‘parts’ of a single AI-feature (for instance, one ‘part’ related to one fictional twin and another ‘part’ related to the other fictional twin).

Yet suppose, as I have suggested, that AI-features do not figure in the identity conditions of ficta. How, then, can one ground the distinctness of indiscernible ficta such as Ménard’s Don Quixote and Cervantes’s Don Quixote? I suggest, first, that the relevant AI-features still fix the identity conditions of such ficta, but only through their results. More precisely, AI-features such as being created by Ménard and being created by Cervantes have two distinct results or products.²² These results are or include two primitively individuated entities, which I shall call ‘traces’. Call the first the ‘PM-trace’ and the second the ‘C-trace’. Traces may be thought of as haecceities, that is, singular features that can characterize only one entity, do not depend for their identity on anything, and at least contribute to grounding the identity and individuation of that entity.²³ Alternatively, traces may be thought of as bare particulars, that is, particulars that metaphysically ‘constitute’ only one ‘thick’ particular (such as Ménard’s Don Quixote), do not depend for their identity on anything, and at least contribute to grounding the identity and individuation of that ‘thick’ particular.²⁴

It is traces – and not the AI-features that produce them – that partake in the identity conditions of ficta. Traces such as the PM-trace and the C-trace thus make Ménard’s Don Quixote distinct from Cervantes’s Don Quixote: the former trace partakes only in the identity conditions of Ménard’s Don Quixote and the latter only in the identity conditions of Cervantes’s Don Quixote. Though depending on AI-features for their production, traces do not include any further entity in their own identity conditions. Thus, they do not depend on the relevant AI-features producing them for their identity, nor do they include such AI-features in their identity conditions (as will become clear, it may be the other way round). Additionally, traces are not identical with the AI-features producing them. In sum, traces do not reduce to AI-features.

Moreover, traces are not identical with the ficta they are traces of, nor do they identity-depend on them (it is the other way round). For example, if traces are bare particulars, they ‘constitute’ ficta. If they are haecceities, they are properties of ficta. In both cases, ficta depend on their own traces for their individuation and identity. Thus, traces do not reduce to the ficta they are traces of.

Mutatis mutandis, traces may also be invoked for fictional twins and fictional members of fictional crowds. In these cases, we may surmise that different AI-features (for example, one for each twin) produce distinct traces, or that different ‘parts’ of a single AI-feature do so. One may even surmise that a single AI-feature – even without having ‘parts’ – produces distinct traces. For example, an author’s creating a pair of fictional twins, or the AI-feature entailed by it, results in two distinct traces. And an author’s creating a fictional crowd, or the AI-feature entailed by it, results in many traces, one for each fictional member of the crowd – though the number of such traces may be indeterminate according to the story.

This latter option is not available to those who invoke AI-features (or ‘parts’ thereof) in the identity conditions of indiscernible ficta. They cannot invoke one and the same AI-feature simpliciter in those conditions; otherwise, distinct yet indiscernible ficta would have the same identity conditions and thus be identical.

Traces may also be invoked to account for the fact that one and the same fictum may have inconsistent features in its identity conditions. It is the single trace figuring in those conditions that makes it the case that there is only one fictum, despite the inconsistency.

Traces are also available to anti-realists. They may hold that distinct ‘surrogates’ of ficta derive from distinct traces produced by AI-features, even if the traces of anti-realists do not individuate ficta but ‘surrogates’ of ficta.²⁵

Finally, traces allow for a new solution to the resurfacing Problem of Circularity discussed above. Recall the problem: some AI-features (such as certain intentions of Zweig) include ficta in their own identity conditions (for example, Dr. B) while, at the same time, fixing or contributing to fixing the identity conditions of those very ficta. This seems to imply, by transitivity, that ficta fix their own identity conditions.

By appealing to traces, one may deny the first conjunct of the first premise of this argument – namely, that the relevant AI-features include ficta in their own identity conditions. One may deny that some intentions of Stefan Zweig that fix, or contribute to fixing, the identity conditions of Dr. B include Dr. B in their own identity conditions. Such intentions may instead include the trace of Dr. B (the B-trace), possibly together with further features (for example, some of those ascribed to Dr. B). In this case, the intention need not have Dr. B in its own content and, therefore, in its own identity conditions. For having an intention concerning the B-trace and the relevant additional features is one thing; having an intention concerning Dr. B is another.

It is one thing that Zweig, when writing The Royal Game, intends to create at least and at most one fictum that is an amateur chess player, persecuted by Nazis, called ‘Dr. B’, and so on. It is another thing that Zweig thinks of Dr. B, that is, the very fictum he has created. In the former case, Zweig intends to create at least and at most one fictum (and this is accounted for through the B-trace) that is such-and-such. The fictum itself is not yet available to him. What is present is only the thought and intention to create something that is such-and-such, an intention whose identity conditions include only the B-trace and the relevant features. In the latter case, by contrast, when Zweig thinks of Dr. B, he thinks of a specific fictum that is already available as a fictum and that he has already created. In sum, it is only the former AI-feature – namely, the intention to create at least and at most one fictum that is such-and-such – that fixes or contributes to fixing the identity conditions of Dr. B.

Three final worries should be dispelled. First, there may still appear to be circularity and reflexivity. Traces are produced by AI-features, such as the intention to create at least and at most one fictum that is such-and-such. Yet those AI-features include those very traces in their own identity conditions. Thus, it may seem that traces fix, or contribute to fixing, their own identity conditions.

This circularity and this reflexivity are only apparent. AI-features (also) depend for their identity upon traces, since they include the latter in their own identity conditions, but they do not depend on traces for their production and existence. Traces do not produce, or contribute to producing, AI-features; rather, traces depend for their production on AI-features. Yet traces do not depend for their identity on AI-features. To produce traces, AI-features need not – and cannot – feature in the identity conditions of traces, so that traces need not identity-depend on AI-features. The ‘cannot’ is justified by the fact that traces have primitive identity conditions and thus do not identity-depend on anything. Finally, ficta depend for their identity on traces, which figure in their identity conditions, but do not do so for their production. At best, ficta depend for their production only on AI-features. Or, alternatively, ficta depend for their identity (but not their production) on traces and traces depend for their production (but not their identity) on AI-features.

At any rate, ficta cannot turn out to depend upon themselves, either for identity or for production – even if both sorts of dependence are taken to be transitive.

Second, recall Horwich and her intention to write a new story about Zweig’s Dr. B. Here it seems that the identity between Zweig’s Dr. B and Horwich’s Dr. B is grounded in Horwich’s thought of Dr. B as an already-created fictum. Does this lead to resurfacing the Problem of Circularity? It does not. There is only one fictum here – namely Dr. B, whose identity conditions include the B-trace and the relevant additional features. These conditions are fixed by Zweig’s intention to create at least and at most one fictum that is such-and-such. That intention, in turn, includes the B-trace and the relevant features in its own identity conditions. Yet Dr. B does not include Horwich’s intention and thought of Dr. B in his own identity conditions. Nor is there any further fictum distinct from Dr. B that includes Horwich’s intention and thought of Dr. B in its identity conditions. At best, one may hold that there is a new ‘appearance’ of Dr. B, in the sense that the very same fictum (namely Zweig’s Dr. B) is ascribed new features in Horwich’s story. This does not affect Dr. B’s identity conditions, which are already fixed. Horwich’s intention and thought of Dr. B only play a role in making it the case that Zweig’s Dr. B is ascribed additional features in her story.

Third and finally, one may reply that introducing traces is a controversial and ad hoc move. Traces conceived as haecceities and bare particulars may well be ontologically controversial, and their admission into one’s ontology requires motivation, as with many other ontological posits. Yet their introduction here is not ad hoc. First, haecceities and bare particulars are independently invoked to account for the metaphysical constitution of ordinary particulars and to distinguish between seemingly indiscernible ones. Second, in tracking the identity conditions of something, at some point or another one should stop: either one accepts entities with primitive identity conditions or one faces circles of dependence or infinite regresses. In this respect, haecceities and bare particulars, precisely because they have primitive identity conditions, are legitimate stopping points when accounting for the identity of ordinary particulars or ficta.²⁶

Competing Interests

The author has no competing interests to declare.