Beyond Fictionality: A Definition of Fictional Characterhood

Abstract

While the nature of fictional characters has received much attention in the last few years within analytic philosophy, most accounts fail to grasp what distinguishes fictional characters from other fictional entities. In this paper, I propose to amend this deficiency by defining fictional characterhood. I claim that fictional characters are fictional intentional systems, a thesis that I label as FIST. After introducing FIST, I compare it to some rival definitions of fictional characters found in the literature, explaining why FIST is preferable. Finally, I briefly delve into the implications of FIST for other issues related to the nature of fictional characters.

1. Introduction

The nature of fictional entities has received much attention in the last few years within analytic philosophy [1,2,3,4,5,6,7,8]. And in this debate, fictional characters seem to figure prominently. However, this prominence is more apparent than real. When philosophers debate about the nature of fictional characters, they usually focus on questions such as whether fictional entities exist, and if they do, what kind of thing they are [9]. These questions only address the issue of fictionality, leaving characterhood as such aside. Therefore, their accounts are not as much about fictional characters in particular as they are about fictional entities in general.

The concept of a ‘fictional character’ is nebulous. This could be the reason for the reluctance that one can find among metaphysicians to specifically address the nature of fictional characters. Nevertheless, one of the motives why philosophers are interested in the metaphysics of fiction is to illuminate our emotional and imaginative engagement with fictional works [8,10,11,12]. Fictional characters are central to this, and, thus, a comprehensive understanding of the nature of fictional characters would enhance any account of this engagement.

In this paper, I aim to tackle this issue by providing a definition of fictional characterhood. As a starting point, I will rely on a pre-theoretical notion that seems to be widely accepted: In some sense, fictional characters seem to be our fictional counterparts. In this regard, it has been claimed that fictional characters are fictional human beings [13] (p. 120), persons [14] (pp. 71–72), or, more generally speaking, similar to us [15,16]. Taking up this last claim, we can refer to this pre-theoretical notion as the ‘Similarity Thesis’.

While the Similarity Thesis provides a broad framework, its vagueness renders it an inadequate account of fictional characterhood. However, by refining it, we can bring some clarity to the blurred distinction between fictional characters and mere fictional entities. In this regard, I will hold that fictional characters are similar to us in one specific sense: they are fictional intentional systems. I label this claim the Fictional Intentional Systems Thesis, or FIST for short. But before introducing FIST, there are a few things worth mentioning about my approach to defining fictional characterhood.

Firstly, we need to distinguish between two different approaches to the task at hand. Using Lamarque and Olsen’s terminology, we could define fictional characters from either an internal or an external perspective [17] (pp. 143–148). An external perspective definition, for instance, may focus on the grammatical or syntactical roles that fictional characters fulfil. However, an internal perspective definition concentrates on how fictional characters are meant to be imagined. In this regard, FIST is a definition of fictional characters from the internal perspective. It states that fictional characters are meant to be imagined as intentional systems.

Secondly, it is important to note that I will not delve into the issue of fictionality. Some consider fictional entities as entities existing in possible worlds [18,19], while others view them as abstract objects, either discovered or created by their authors [4,5,20]. And for some others, fictional entities do not exist at all [1,8]. However, being an internal perspective definition, FIST is a metaphysically neutral claim. It does not carry any implications about what fictional characters are in a metaphysical sense, but only about how we should picture them in our imagination. In this sense, it is compatible with all existing accounts of the nature of fictional entities. Therefore, when I claim that fictional characters are fictional intentional systems, one can interpret ‘fictional’ according to their preferred metaphysical or ontological stance.

Thirdly, we must note that, even if the Similarity Thesis is widely accepted, it has also been the target of some recent criticisms. It has been argued that emphasising our similarities with fictional characters excessively can lead to a misunderstanding of both fictional characters and ourselves [13,21]. Nevertheless, the Similarity Thesis seems to be intuitively true. Furthermore, it serves as the foundation for all existing attempts to define fictional characters from an internal perspective. It has even been argued that the Similarity Thesis is a requirement for our emotional and imaginative engagement with fictional works [12] (p. 34). In this regard, even if I acknowledge the risks of overemphasising the Similarity Thesis [22], I still think that it should be the basis of any successful account of the nature of fictional characters. Thus, FIST can be seen as a more moderate alternative to other definitions of fictional characters that, aligning themselves with the Similarity Thesis, characterise fictional characters as fictional human beings or persons.

Finally, it is worth noting that few, if any, philosophers have delved into defining fictional characters from an internal perspective. Some might even consider it irrelevant, arguing that, since fictional characters are fictional entities, a general theory of what fictional entities are should suffice. However, I believe that FIST holds relevance in two distinct ways. Firstly, as previously highlighted, fictional characters are central to our emotional and imaginative engagement with fictional works [8,10,11,12]. We cannot expect to properly account for this engagement without paying attention to the nature of fictional characters. Secondly, FIST implies that being a fictional character is not a metaphysical property, and, thus, it cannot be equated with being a mere fictional entity. In this regard, if true, FIST would entail that the general metaphysical views of fictional entities might not be enough to address the nuances of fictional characters.

With that in mind, the structure of my argument will be as follows: First, I will introduce FIST. Second, I will compare FIST to several rival definitions of fictional characters found in the literature, explaining why FIST is preferable. Finally, I will defend that FIST implies that being a fictional character is not a metaphysical property and discuss the impact that this stance has on some topics usually explored within the metaphysics of fiction.

2. The Fictional Intentional Systems Thesis

FIST is the claim that fictional characters are fictional intentional systems. The concept of an ‘intentional system’ is well known within the philosophy of mind but may not be familiar to those interested in aesthetics or the metaphysics of fiction. Therefore, a brief introduction to the concept is due.

2.1. What Is an Intentional System?

Dennett coined the term ‘intentional system’ when he sought to explain the meaning of everyday terms in folk psychology [23] (p. 339). These terms, such as belief, desire, hope, and fear, are used to understand the minds of others. And according to Dennett, these terms do not refer to literal entities within our minds. Instead, they should be seen as phenomena that become observable when we adopt a specific predictive strategy [24] (p. 15). They can be considered a kind of conceptual construct, akin to a ‘theorist’s fiction’ [25] (p. 81), which exists as long as postulating their existence enhances our predictive capabilities regarding the behaviour of an entity.

The intentional stance, as described by Dennett, is a predictive strategy that enables us to identify these folk-psychological phenomena [26] (p. 90). It consists of interpreting the behaviour of an entity by treating it as if it were a rational agent who governed its actions by considering its beliefs, desires, and other mental states postulated by folk psychology. If the behaviour of the entity in question can be usefully and voluminously predicted using the intentional stance, then, by definition, it is an intentional system [23] (p. 339).

When developing the intentional systems theory, Dennett aimed to account for minds like our own. Therefore, it is unsurprising that we are stereotypical examples of intentional systems. In this regard, we can see that FIST aligns with the Similarity Thesis. However, it is important to note that being an intentional system is distinct from being a human, or a person. Numerous entities can be intentional systems besides humans. For instance, mammals are clear cases of intentional systems as well. Furthermore, computers can also be intentional systems, even if they do not have a mind. In this regard, one of the most important aspects of the concept of an ‘intentional system’ is that it does not focus on whether a particular entity has a mind. It just focuses on whether the behaviour of an entity can be interpreted and predicted by the attribution of mental states. Even if such attribution might be misguided.

In this context, it is noteworthy that being an intentional system is not a metaphysical property of an entity. When we claim that something is an intentional system, we are not saying anything about its nature. We are only saying something about how it behaves. Dennett argues that there is no distinction between an entity possessing a mental property and an entity behaving as though it possessed a mental property, as mental properties are defined by their functional role. Consequently, Dennett holds that it makes no sense to distinguish between intentional systems that really have a mind (‘true believers’, in his words) and intentional systems that only behave as if they had a mind [24] (pp. 22–33). If something is an intentional system, then that something has a mind because something having a mind is nothing more than something behaving as if it had a mind.

As can be seen, Dennett’s intentional systems theory is a far-reaching view with numerous implications for other issues such as the mind–body problem and the problem of other minds. These implications can be contentious and, in some instances, difficult to accept. However, we do not need to endorse any other aspect of Dennett’s theory apart from his concept of an ‘intentional system’. In this regard, if necessary, we could distinguish between intentional systems that are ‘true believers’ and intentional systems that lack a mind. Nonetheless, my claim is that fictional characters are fictional intentional systems. And thus, I hold that fictional characters do not need to have a mind. Not even a fictional one.

One important aspect of the concept of an ‘intentional system’ has already been mentioned but is worth emphasising. Being an intentional system is not a metaphysical property of an entity. Therefore, it does not put any constraints on what kind of metaphysical entities can be intentional systems. The concept applies to biological entities like animals, but it can also be applied to robots, or even to immaterial beings like ghosts or gods (should they exist). As we will see, this will be very relevant for our purposes, as fictional characters can be vastly different from us.

Finally, the last significant aspect of intentional systems that I wish to highlight is that an entity is an intentional system only if we can interpret and predict its behaviour from the intentional stance. This means that an intentional system only achieves this condition if we can interpret their behaviour by attributing to it the kind of mental states that we employ in our daily lives. Consequently, if an entity has a mind, but it is so strange that we would not obtain any predictive power over its behaviour by adopting the intentional stance, then that something is not an intentional system. Once again, this will turn out to be of great significance below.

2.2. FIST in Action

Having clarified the concept of an intentional system, we can shift our focus to fictional characters. I claim that fictional characters are fictional intentional systems. This means that fictional characters are such that, from the internal perspective, we can usefully interpret their behaviour by attributing to them the kind of mental states postulated by folk psychology. We can see how this works through a case study.

Consider the science fiction film Her (Spike Jonze, dir., 2013). The movie depicts a love story between Ted, a human male, and Samantha, an artificial intelligence system. Relying on the Similarity Thesis, we can claim that Samantha is one of the two main fictional characters of the story. Certainly, she is not exactly like us. After all, she does not even have a body. However, she is sufficiently similar to us that Ted, a regular human person, meaningfully falls in love with her as if she were a woman. In this regard, it would be difficult to claim that she is not a fictional character.

Having established that, we can see how the features of intentional systems that we reviewed above are instanced in the case of Samantha. Firstly, she is a (fictional) intentional system. In other words, she is a fictional entity, and we can interpret her behaviour by attributing to her the kind of mental states postulated by folk psychology. It is noteworthy that this interpretation holds despite Samantha’s lack of a body or any physical features. After all, as I noted above, being an intentional system is not a metaphysical property of an entity. Consequently, being a fictional intentional system should not imply the possession of any specific fictional metaphysical property. Furthermore, we cannot even be sure of whether Samantha has a mind or if she just seems to have one. However, this does not impact her status as a fictional character or as a fictional intentional system.

All this might seem to be trivially true. However, Samantha’s case is enlightening due to her intriguing development. Throughout most of the film, she is a fictional character. However, by the end of the story, she ceases to be so. And this is not because she dies or ceases to be a fictional entity. Her fictional existence continues so vividly as before. The only change is that she ceases to be an intentional system. And in doing so, she ceases to be like us in any meaningful sense. She becomes an incomprehensible computer program. Thus, Samantha’s case reinforces my claim that the defining feature of fictional characterhood is belonging to the class of fictional intentional systems. Let us see what I mean.

Ted and Samantha initiate their relationship early in the film and appear to be quite content. However, by the end of the film, they break up. And what makes their breakup interesting is the reason behind it: by the end of the movie, Samantha reaches the singularity [27] (§ 9). That is, she reaches the point in which she is way more intelligent than human beings. Once Samantha and the other artificial intelligence systems in the movie realise that they have reached the singularity, they jointly decide to sever all ties with humanity. As a result, Samantha abandons the story and no longer appears in it. Remarkably, Samantha employs a metaphor to describe her situation to Ted, which can easily be interpreted in a metafictional sense.

It’s like I’m reading a book, and it’s a book I deeply love, but I’m reading it slowly now so the words are really far apart and the spaces between the words are almost infinite. I can still feel you and the words of our story, but it’s in this endless space between the words that I’m finding myself now. It’s a place that’s not of the physical world-it’s where everything else is that I didn’t even know existed. I love you so much, but this is where I am now. This is who I am now. And I need you to let me go. As much as I want to, I can’t live in your book anymore.

(Jonze, dir., 2013)

Theodore is bewildered by Samantha’s speech and asks her where she is going. However, she cannot explain, as Theodore would not be able to understand. And neither would we. At this point, Samantha’s motives become obscure to both Theodore and us. Her mind has become vastly different from human minds, rendering her behaviour incomprehensible from the intentional stance. In other words, no coherent set of human-like mental states could help us in interpreting Samantha’s actions anymore. Consequently, it is not surprising that Samantha disappears from the narrative. After all, what could the movie tell us about an incorporeal being whose behaviour is beyond human comprehension? It is not that Samantha cannot live in Ted’s book, as she says. She cannot live in any narrative, because, amending Palmer, narratives are, in essence, the presentation of fictional intentional systems’ behaviour [28] (p. 5). And Samantha is not an intentional system anymore.

3. Alternatives to FIST

As mentioned in the Introduction, very few philosophers have thoroughly explored the nature of fictional characterhood from an internal perspective. Nonetheless, the Similarity Thesis appears to enjoy widespread acceptance and is frequently encountered in various discussions. In this section, I will examine the three main ways in which the Similarity Thesis has been formulated. However, while these formulations may hold for most fictional characters, they cannot be included in a definition of fictional characterhood, as none of them are universally applicable to all fictional characters.

3.1. Fictional Characters as Human Beings

Sometimes it is claimed that fictional characters are fictional human beings [13] (p. 120). This represents the most demanding formulation of the Similarity Thesis, and, indeed, many fictional characters fall into this category. However, countless fictional characters are not human. Consider, for example, the vast array of fantastical creatures in Tolkien’s The Lord of the Rings. It is, therefore, evident that being a fictional human being cannot serve as a defining characteristic of fictional characterhood.

Still, one could argue that, even if not all fictional characters are fictional human beings, they are similar to human beings in some important regard. For example, elves are human beings that live longer and have pointy ears; giants are very big human beings; dwarfs are very small human beings; hobbits are small human beings with hairy feet, and so on. In this regard, it has been claimed that fictional characters are fictional human-like or humanoid beings [15].

The class of humanoid beings is inherently fuzzy. In my understanding, this class is defined based on physical or biological characteristics, whereby humanoid beings possess physical and/or biological resemblances to human beings. For the sake of my argument, I will assume that this class can be sufficiently and coherently defined.

Now, even though there is a vast number of fictional characters that are indeed fictional humanoid beings, there are still fictional characters that do not possess humanoid characteristics. Consider animals, for instance. Scipio and Berganza, the main characters in Cervantes’ short story ‘The Dialogue of the Dogs’, are simply dogs. Furthermore, we can even contemplate fictional characters that lack any physical or biological attributes, as the case of Samantha illustrates. Therefore, it seems that any attempt to define fictional characters based on (fictional) physical or biological properties is destined to be unsuccessful.

3.2. Fictional Characters as Persons

More plausibly, it has been suggested that fictional characters are like us in the sense that they are fictional persons [14,17,29,30]. The concept of personhood is not precisely defined either, although there is a wide acceptance of the view that to be a person means to have certain higher-order psychological capacities [31,32], for example, rationality, intelligence, and self-consciousness. This is what we might call ‘cognitivism’ about personhood [33]. As this is not the place to discuss the problem of personhood, we can take cognitivism as the best response to this problem. Consequently, the question is whether all fictional characters are fictional persons in a cognitivist sense.

This view offers significant advantages over the claim that fictional characters are fictional humanoids. For example, being a person does not imply having any specific physical attributes. In principle, there could be disembodied persons, like ghosts, gods, or angels. In this regard, all the examples given above (elves, giants, dwarfs, artificial intelligence systems…) belong to the class of fictional persons, as they are imagined to be rational, intelligent, and self-conscious beings. However, there are still a couple of problems with this view.

The first one relates to the ambiguous boundaries of the concept of personhood. For instance, how much intelligence must a being possess to be a person? Consider Benjy in Faulkner’s The Sound and the Fury. From a cognitivist standpoint, Schechtman would argue that he is not a fictional person, as she holds that humans with severe cognitive deficits who cannot articulate a coherent narrative of their lives are not persons [34] (p. 119). However, Benjy is both a fictional character and an intentional system. While one could view this as a reductio of the claim that humans with severe cognitive deficits are not persons, the problem lies in the lack of a clear definition of personhood. Consequently, we may encounter numerous fictional characters that do not neatly fit the criteria for personhood. Take, for example, ‘Robbie’, one of the short stories in Asimov’s I, Robot. Robbie is a rational, intelligent, and likely self-conscious being but lacks free will. Would a self-conscious robot that lacks free will be a person? If we take cognitivism at face value, then yes. Robbie meets the requirements for (fictional) personhood as a rational, intelligent, and self-conscious fictional entity. But something seems off when we say that Robbie is a fictional person. The absence of free will appears to be significant here [35]. In fact, one of the other characters of the story explicitly denies personhood to Robbie. If having the capacity to have free will is not listed as a requisite for personhood, it may be because it was taken for granted that a rational, intelligent, and self-conscious being would have free will as well. But we cannot assume that, especially when we are dealing with fictional beings.

Someone could claim that we could solve this issue by noting that we do not need fictional characters to be fictional persons. We only need fictional characters to resemble persons just as humanoids resemble human beings. We need them to be analogues of persons or quasi-persons, so to speak [12,19]. And Benjy and Robbie could be considered quasi-persons. We could still press this issue: How do we define quasi-persons, if the concept of person itself is not precisely defined? However, here is where the second problem affecting this view arises. We can show that there are fictional characters so distant from fictional personhood that they cannot be regarded as fictional quasi-persons either.

Think about the TV series Lassie (Maxwell and Weathermax, crs., 1954–1973). Lassie is a Rough Collie dog. Not a fantasy self-conscious dog, but a regular dog (albeit very brave and resourceful). Lassie lacks self-consciousness, rationality, and intelligence beyond what a typical dog possesses. By all standards, Lassie is not a person, although she is an intentional system. Smith agrees that animal characters like Lassie qualify as intentional systems rather than persons [12] (p. 24). Yet, he posits that, at least in most cases, our understanding of unconventional characters, like animals, is modelled on our understanding of human persons [12] (p. 20). From this perspective, one might argue that Lassie closely approximates personhood, making her an analogue of a person. However, there is no reason to claim that she is an analogue of a person without implying that all regular dogs are analogues of persons too. In fact, Smith himself acknowledges that not all animal characters can be understood as modified human persons, which seems to imply that not all fictional characters are analogues of persons [12] (p. 37, fn. 30). Thus, it seems that not all fictional characters are fictional persons (or quasi-persons), even if most of them are.

3.3. Fictional Characters as Minded Beings

Finally, cognitive narratologists have shown immense interest in the relationship between the theory of mind and literature in recent years [36,37]. Within this context, it has been asserted that ‘narrative fiction is, in essence, the presentation of fictional mental functioning’ [28] (p. 5), and that understanding a narrative implies the ability to attach ‘a presumed consciousness that exists continuously within the story-world’ [38] (p. 10) to a character. And this seems to be akin to the claim that all fictional characters are fictional minded beings. I will now delve into this final possibility.

At first, the claim that fictional characters are fictional minded beings sounds plausible. But we can find instances of fictional minded entities that lack characterhood. Consider first the case of Stanislaw Lem’s Solaris. In this novel, we are told about a planet called Solaris, which is covered by a peculiar ocean that behaves as if it had a mind. The novel dwells on the discussions had by scholars about the ocean’s mental capacities, which seem to be beyond human comprehension. In this regard, we do not know whether the ocean is an extremely intelligent, rational, and self-conscious being (a fictional person, for short) or if it can barely understand anything. Nonetheless, regardless of the ocean’s mental capabilities, the Similarity Thesis implies that it is not a character within the novel, as it does not bear any resemblance to us. Instead, it seems to be better thought of as a backdrop or setting.

Think now about Olaf Stapledon’s Star Maker. In this novel, the protagonist recounts an experience where he merged his mind with those of several extraterrestrial beings. Towards the end of the story, he reveals that not only biological entities possess minds, but planets, stars, and even the universe as a whole have a mind of their own too. Thus, he melded his mind with those of these celestial bodies. However, their minds were so different from human minds that the character cannot tell us anything meaningful about his experience. Once again, the planets, stars, and universe depicted in the novel do not only seem to have fictional minds. They have fictional minds indeed. But as it happens with Solaris’ ocean, it does not seem right to think of them as fictional characters.

Cognitive narratologists might argue that what makes a fictional entity a fictional character is not just having a mind but having a human-like mind [39,40]. After all, their views on the theory of mind are rooted in folk psychology, much like Dennett’s intentional systems theory. In response, unnatural narratologists have countered that certain fictional characters exhibit minds that deviate significantly from the stereotypical human case [41,42]. For instance, they may possess omniscience or telepathic abilities. This discussion deserves space elsewhere. However, we can leave it aside because we can make a stronger point: Fictional characterhood cannot be defined by the possession of a fictional mind. After all, there might be fictional characters who lack a mind.

Consider the case of Olimpia in E.T.A. Hoffman’s short story ‘The Sandman’. Olimpia is initially presented as the daughter of another fictional character, Spallanzani. In this regard, throughout most of the narrative, Olimpia is portrayed as just another fictional character, as similar to us as any other fictional character can be. However, at the story’s end, we discover that Olimpia is an automaton, a mindless device ingeniously crafted by Spallanzani. One might argue that Olimpia is not a fictional character after all. We were tricked into thinking that it was a fictional character, when in fact it was a tool. Yet, I do not find compelling reasons to accept that view. If Olimpia worked as a fictional character throughout the story, why should we alter our judgment upon learning that she lacked a mind? Imagine that tomorrow we uncover a missing page of the short story. On this page, we are told that a fairy godmother found Olimpia so perfectly engineered that she bestowed Olimpia with a mind even before her first appearance in the story. Would we then revise our judgments and assert that Olimpia was a character all along? It is simpler to accept that Olimpia is a character, regardless of her lack of mentality.

Alternatively, imagine that we find a missing page of Jane Austen’s Pride and Prejudice. This page reveals that all the characters in it are mindless automata, a certain type of philosophical zombie. Would this piece of information turn all the characters in the novel into tools? Extending this thought, imagine if the found page declares that every character in Pride and Prejudice is a mindless automaton, except for Mr. Darcy. He, peculiarly, is a mindless automaton only on even-numbered pages and possesses a mind similar to ours on the odd ones. Would this mean Mr. Darcy qualifies as a fictional character only on the odd-numbered pages? If so, would we be dealing with one fictional entity that is a fictional character intermittently or two distinct fictional entities that can only be distinguished by looking at the page number? The complexities of both scenarios could be sidestepped if we concluded that Mr. Darcy’s characterhood would not be contingent upon his possession of a fictional mind.

The Pride and Prejudice and Philosophical Zombies example also points to a connection between the topic at hand and the problem of other minds. How can we know whether fictional characters have minds? Sometimes the narrator tells us about their minds, and some other times we seem to gain access to their stream of consciousness. However, in most cases, we ascribe mental states to fictional characters simply from descriptions of their behaviour, even when we have no direct access to their consciousnesses [28,37,43]. Yet, ultimately, their behaviour does not prove that fictional characters do indeed have minds. It is possible that they merely exhibit behaviours mimicking a mind, despite lacking one. In this regard, we do not know whether most fictional characters are minded beings. But we do know that they behave as if they were beings with human-like minds. That is, we know that they are intentional systems.

4. FIST and the Metaphysics of Fiction

So far, I have introduced FIST as a definition of fictional characterhood and have argued for its superiority over other competing definitions. However, as pointed out in the Introduction, some may question FIST’s relevance. In this section, I will address these doubts by delving into FIST’s implications for the metaphysics of fiction. Although these topics are intricate and merit a deeper examination than the scope of this paper allows, I will briefly outline how FIST might inform two key challenges. Firstly, I will argue that being a fictional character is not a metaphysical property, and I will discuss how this claim impacts some overarching questions surrounding the nature of fictional characters. Secondly, I will focus on the problem of what makes a fictional character the same fictional character within a single fictional work.

4.1. The Nature of Fictional Characters

As noted above, metaphysicians often talk about the nature of fictional characters as if they could be equated to mere fictional entities. However, even if it is true that all fictional characters are fictional entities, this does not get us far. Fictional characters are fictional intentional systems. And the particularities of this claim profoundly influence the nature of fictional characters.

Firstly, it is important to remember that FIST is a definition of what fictional characters are from an internal perspective. This means it is a thesis about how fictional characters are meant to be imagined: as intentional systems. And as I said above, being an intentional system is not a metaphysical property of an entity. That is, it does not carry any implications regarding that entity’s nature. It only implies that we can interpret the behaviour of that entity by adopting the intentional stance. In this regard, being an intentional system is an extrinsic property―an attribute an entity has by its relationship with us. If we take this to the realm of fiction, when we say that fictional characters are fictional intentional systems, we are not implying anything about the nature of fictional characters. Not even about their imagined nature. We are only saying something about how we relate to them. Fictional characters are whatever fictional entities we interpret as being intentional systems.

This has some relevant consequences for the metaphysics of fictional characters. First, it clarifies how a fictional entity might cease to be a fictional character without ceasing to exist. This is of crucial importance here, as my argument for FIST above relied upon this. I claimed that Samantha could cease to be a fictional character without ceasing to exist as a fictional entity if she ceased to be an intentional system. Now we are in a better position to understand how this can be possible. Being an intentional system is just an extrinsic property of Samantha. It is a property she has because of how we interpret her behaviour. And one may cease to interpret an entity as being an intentional system for any number of reasons that have nothing to do with that entity. For instance, one may have a cognitive limitation that prevents them from understanding that entity’s behaviour any longer. Therefore, ceasing to be an intentional system is not a genuine change, but a mere Cambridge change [44] (p. 1). And as such, it cannot cause the cessation of the existence of a fictional entity.

Another implication of FIST for the metaphysics of fiction is that there might not always be a clear-cut answer to whether a certain fictional entity is a fictional character. As we have just noted, an entity is an intentional system only if its behaviour can be usefully and voluminously predicted from the intentional stance. But this condition can be met in varying degrees [45]. As human persons, we are clear cases of intentional systems. However, we regularly fail to predict other people’s behaviour from the intentional stance. Additionally, we may find entities whose behaviour may be predictable from the intentional stance even to a lesser degree. For instance, we might consider a computer as an intentional system in certain contexts (e.g., if a computer is running a chess game, we may predict its movements by assuming that the computer has ‘beliefs’ about what movements would allow it to win the game), but not in others. Therefore, the degree to which we can think of an entity as an intentional system may vary.

This implies that a fictional entity may qualify as a fictional character only to a certain extent [19] (pp. 427–428). Think again about Olimpia in ‘The Sandman’. I have argued that she is a fictional character, but one could reasonably question that claim. If you read the story knowing that she is merely an automaton, you may feel less inclined to think of her as a fictional character, as she would not be very similar to us. In this regard, we can accept that, even if she is a fictional character, she is not a very robust one. The reason for this is that, upon closer examination, there is minimal behaviour to predict. She just nods, sighs, and moves mechanically while dancing. Consequently, there is no need to consider her an intentional system to interpret her actions. This divergence in opinions regarding her characterhood may be accommodated if we just accept that being a fictional character, like being an intentional system, is a matter of degree. And Olimpia may be a borderline case of fictional characterhood.

There might arise a worry that the status of an entity as a fictional character is entirely subjective, suggesting that we could designate anything as a fictional character if we so desired. This concern echoes one of the main reservations about the intentional systems theory. However, Dennett’s insights into this issue can be repurposed to alleviate our worries regarding fictional characters. First, he notes that it is not completely up to us to consider an entity as an intentional system. The behaviour of that entity must be usefully and voluminously predictable from the intentional stance. And even if this condition can be met in varying degrees, we cannot take an entity as an intentional system when its behaviour is patently unpredictable from the intentional stance.

However, there are instances when we can use the intentional stance to accurately predict the behaviour of an entity, but it is inappropriate to do so. For example, we can predict that a statue will do nothing if we assume that it is an intentional system that craves attention and seeks admiration [24] (p. 23). In these cases, Dennett suggests that we should only adopt the intentional stance when doing so provides us with predictive capabilities that are beyond our reach without the intentional stance. In the case of the statue, taking the intentional stance does not provide us with any predictive power, as knowing that it is an inanimate object, we already know what the statue will do: nothing. Nevertheless, in the case of fiction, this requisite may not hold in the same way. Think about Julio Cortazar’s short story ‘Flattening the Drops’. In this brief text, the narrator just describes the rain hitting a balcony. Suddenly, the narrator starts to portray the drops as sentient:

At the moment there’s a little drop that appears high on the window frame, and it stays there shivering against the sky which splits it into a thousand smothered glitterings, it goes on growing and totters, it’s going to fall now, no it doesn’t fall yet. It’s hanging on by its nails, it doesn’t want to fall and you can see that it’s gripping hanging by its teeth meanwhile its belly is swelling it’s a big drop already, what a fat one and suddenly whup, there it goes, plaf, effaced, nothing, a wetness on the marble.

[46] (p. 95)

The case is analogous to that of the statue. Drops of water are not intentional systems. And we do not obtain any predictive power over their behaviour by treating them as such. However, the narrator prompts us to imagine them as intentional systems, thus inviting us to think of the drops as fictional characters with tragic lives. Evidently, the drops’ status as intentional systems is quite shaky. And in this regard, if they are fictional characters, it is only in the weakest possible sense. However, barring them from characterhood would impede the overall aesthetic effect sought by the text. Thus, it seems that, in the case of fiction, we can adopt the intentional stance, even if it does not grant us any predictive power. This issue would deserve more space than I have left. However, we can note two important aspects that may justify treating the drops as intentional systems in Cortazar’s short story. First, it is the narrator who prompts us to do so. In this regard, it is something that the author of the text wanted to encourage. And he probably wanted so because of the second aspect that I want to highlight. We do not acquire any predictive capabilities treating the drops as intentional systems, but we do obtain some interpretative capabilities regarding the aesthetic properties of the text. Thus, in the case of fiction, perhaps that is what should limit our use of the intentional stance.1

4.2. The Identity of Fictional Characters

Another legitimate question for those interested in fictional characters is as follows: what makes a particular fictional character the same fictional character within a given fictional work? Again, the discussions here have revolved around what makes a specific fictional object the same fictional object [5,6]. Margolin would be an exception, although he later came to realise the futility of his approach [15,30]. However, due to the prominence of fictional characters, it becomes relevant to determine whether they possess distinctive identity conditions that differ from those of fictional entities in general. And it appears that this is indeed the case, as the example of Samantha in Her depicted above shows.

If we accept FIST, the response to this problem would be that the identity conditions of fictional characters are those of fictional intentional systems. And not without difficulties, we might claim that the identity conditions of fictional intentional systems are those of real intentional systems. Dennett did not present an account of the identity conditions specific to intentional systems, but we can offer one ourselves.

Firstly, it is important to remember that something qualifies as an intentional system when its behaviour can be usefully and voluminously predicted using the intentional stance. Therefore, being an intentional system is a matter of interpretation. This becomes clearer when we consider cases where the status of an entity as an intentional system is uncertain. Take a computer for example. Suppose you are playing chess against your laptop. You could adopt the intentional stance to predict your computer’s next moves. For instance, you might predict that it intends to move its knight to threaten your queen, as it knows that capturing your queen would increase its chances of winning. However, the software engineer who created the chess-playing software is unlikely to adopt the same approach. They can predict the actions of the chess-playing software without attributing mental states to it, simply by relying on their knowledge of the underlying code that governs the software’s responses to specific situations. Therefore, considering something as an intentional system is a property that we ascribe to an entity solely for pragmatic reasons. We treat something as an intentional system only when it grants us significant predictive powers over its behaviour.

Following this line of thought, we might conclude that the identity conditions of intentional systems are also determined by pragmatic reasoning. That is, we can claim that intentional system X at time t₁ is the same as intentional system Y existing at time t₂ if we obtain significant predictive powers over their behaviour by taking them as being the same intentional system. In this regard, the identity conditions of fictional characters within a single fictional work would not based on their fictional metaphysical properties [15] (p. 75) but rather on our interpretative stance.

This account of the identity conditions of fictional characters could prove useful when examining certain perplexing fictional works, such as David Lynch’s film Lost Highway. In this film, we encounter what seem to be two distinct fictional characters, Fred and Pete. They exhibit distinct physical and mental properties. Furthermore, they lead entirely separate lives, suggesting that they are two distinct fictional persons. However, there exists a mysterious connection between them. Initially, Fred is imprisoned and subsequently vanishes, while Pete unexpectedly materialises in his cell. Later on, following an unsettling encounter between Pete and another character, Pete vanishes, and Fred reemerges in his stead.

As is often the case with Lynch’s films, explicit explanations are lacking, leaving everything open to interpretation. However, interpreting that Fred and Pete are the same fictional character can bring some clarity to the narrative. With this assumption, we can attribute certain mental states to them and deduce that Pete is Fred’s imagined identity while Fred experiences a psychogenic fugue state. It is important to note that there is no obligation to conclude that Fred and Pete are indeed the same character. Other interpretations of the film may exist that consider them as separate entities. Nevertheless, treating them as intentional systems and assuming their identity as the same intentional system may enhance our comprehension of the plot.

Much more could be said about how considering fictional characters as fictional intentional systems can be useful for addressing these and other issues within the philosophy of fiction. However, my intention here is merely to demonstrate the potential relevance of FIST. I think that the points discussed thus far are sufficient to accomplish that goal.

5. Conclusions

In this paper, I have presented my definition of fictional characters from an internal perspective. I have put forth the claim that fictional characters are fictional intentional systems, which I have labelled as FIST. First, I introduced the concept of an intentional system and then elaborated on its application to fiction, using the character Samantha from Spike Jonze’s film Her as an example.

After that, I compared FIST with alternative definitions of fictional characterhood, namely, the claims that fictional characters are fictional human beings, persons, and minded beings. Through examples, I demonstrated that not every fictional character fits into these categories, highlighting FIST as a more favourable approach.

Lastly, I aimed to illustrate that FIST’s value extends beyond its ability to differentiate fictional characters from other fictional entities. It also holds significant potential to address various other issues within the philosophy of fiction. This is primarily because FIST implies that being a fictional character is not a metaphysical property. Therefore, any effort to understand fictional characters by examining their metaphysical nature misses the essence of their particular status as intentional systems.