1. Introduction

1Emotions and judgements sometimes come into conflict, as the phenomenon of “recalcitrant emotions” attests. For instance, someone might fear a harmless spider even while knowing it poses no danger or might feel anger toward a partner for actions that they recognize occurred only in a dream. These cases impose normative pressure: we ought not feel an emotion that conflicts with our considered judgements. Indeed it seems to be a (defeasible) rational imperative to resolve the conflict.

2The same kind of conflict between emotions and judgements appears to arise in our engagement with works of fiction. We judge that Anna Karenina does not exist and therefore has not suffered, yet we feel pity for her in reading Tolstoy’s novel. We know that Abraham Lincoln never fought hordes of the undead, yet we root for his victory against them in Seth Grahame-Smith’s Abraham Lincoln: Vampire Hunter. However, by contrast with cases of recalcitrance we experience no normative pressure: there seems to be no problem with feeling emotions toward fiction that conflicts with our considered judgements. There is no rational imperative to resolve the conflict.

3Why is there a normative difference between recalcitrant and fiction-directed emotions? This is the question we address in this paper.

2. Puzzles

5The puzzle raised by recalcitrance, then, is to explain why and how emotions generate normative demands weaker than beliefs or judgements but stronger than perception. As Döring (2008) puts it, we need an account of “conflict without contradiction”.

7Colin Radford (1975) famously argued that our emotional responses to fictional characters and events are “inconsistent and incoherent”, thus irrational. Although commentators have often interpreted Radford as committed to a cognitivist theory of emotion, closer attention to his arguments (in the original paper and in numerous replies to critics) demonstrates that he rejects that position; instead, he argues that in responding to fictional characters we knowingly violate a norm that governs emotions (Friend 2020). His argument in a nutshell is this: In other contexts, fearing that which we judge to be harmless or being angry at someone we know not to have done anything to deserve it would be cognitively irrational, insofar as we know that the emotion is unfitting (Robinson 2005; Scarantino, de Sousa 2018). Since (as Radford argues at length) there is no relevant difference in kind between emotions toward fiction and emotions in other contexts, they are subject to the same normative constraints. Therefore, if emotions in other contexts are cognitively irrational when they conflict with considered judgements, so too must be emotions toward fiction.

8In other words, according to Radford emotions toward fiction are recalcitrant emotions: they persist even when we judge that their object lacks the relevant properties, and even that it does not exist. As we have seen, however, this apparent conflict seems to generate no normative pressure for resolution. We do not appear to be violating a norm in emotionally responding to fictional characters and events. But the structure of the predicament looks to be just the same as in the non-fiction-involving cases, so it’s surprising that there is an intuitive normative difference.

13The alternative explanations assume that our attitudes in the fiction context remain ordinary judgement and emotion, but that the norms governing one or the other attitude are changed. According to (ii) it is the norms governing judgement (or temporary belief; see e.g. Smuts) which are affected. If these judgements were appropriately constrained by engagement with fiction rather than reality, they would, according to the suggestion on offer, be subject to a norm of fictional truth, rather than truth. In that case, the judgement that Anna Karenina is a human being who suffers would be apt. This move seems to dissolve any normative tension: pitying Anna when you believe that she has suffered looks perfectly appropriate. So you aren’t guaranteed to either violate the norm governing the emotion or the belief/judgement; the judgement is off the hook.

14But this is not a fully satisfactory response. It is the pity which appears to violate the fittingness norm and thereby to generate the trouble. Whatever our temporary “beliefs”, it remains the case that Anna does not really exist and has not really suffered. So, although a shift from truth to fictional truth can alleviate the pairwise guarantee of normative conflict, it fails to explain why our pity seems apt.

3. Cancellation contexts

16One possibility is that the normative difference stems from a shift in context, and more specifically, from a context-sensitive change in the norms that govern emotion. The proposal would be that in the fiction case, we are in a kind of “cancellation context” for emotional norms.

17The notion of a cancellation context is deployed by Peter Hanks (2015) in his argument that propositions are acts of predication, such as predicating “being yellow” of a tennis ball. Acts of predication, on Hanks’s view, constitutively bring with them a commitment to the truth: e.g., in predicating yellowness of the ball, one incurs a commitment to its being the case that the ball is yellow. Hanks (2015; 2016) notes, however, that there are contexts – he mentions the antecedents of conditionals, fiction, pretence, and quotation – in which one can perform an assertoric act type without incurring the usual commitments. He maintains that these are cancellation contexts: contexts that suspend or cancel the normative commitments typically associated with a given act.

18Although Hanks’s overall account of propositions and predication is controversial (see e.g. García-Carpintero forthcoming), we will not enter that controversy here. Our interest is solely in the concept of a cancellation context. Hanks illustrates it in a way particularly relevant to our discussion:

Consider an actor onstage, uttering lines during the performance of a play. The actor is uttering sentences of English with their usual meanings, but these utterances do not count as assertions. We do not hold the actor accountable for any of the things she says during the play. The character in the fictional world of the play might be asserting these sentences, but the actor herself, the non-fictional person, is not. This is because the actor is uttering these lines in a context in which uttering a declarative sentence with its usual meaning does not count as the performance of an assertion. This is a good example of cancelled predication. Being onstage while acting in a play is a cancellation context, and the acts of predication one performs in that context are acts of cancelled predication. … Onstage or off, by uttering “a is F” the actor predicates being F of a. Offstage, this act of predication is an assertion. Onstage it is not (2016, 1390).

19The example suggests that fiction can function as a cancellation context. For Hanks, acts of predication in the context of performing a play still count as genuine acts of predication; but the assertive force, and correlative norms of responsibility, are cancelled. The corresponding proposal would be that in the context of engaging with a work of fiction, we experience emotions of the usual kind, but the norms governing them are cancelled. To feel fear toward a fictional threat or pity toward a non-existent person is not to violate norms of fittingness or correctness, because in the fictional context, those norms do not apply as usual.

21This proposal allows us to take the apparent normative difference between recalcitrant emotions and emotions toward fiction at face value. It preserves the claim that our emotional responses to fiction are genuine, affectively rich, and evaluable – but evaluable by different standards. Emotions towards fiction are the same as emotions toward reality; they simply occur in a context in which Correctness is cancelled.

22Despite these virtues, the proposal faces a serious challenge: that Correctness is not always cancelled in the context of engaging with fiction (Friend 2020). In reading Chimamanda Ngozi Adichie’s Half of a Yellow Sun, I might learn about the atrocities committed during the Nigerian Civil War. If I consequently pity the real victims of the conflict, this emotion will be subject to Correctness. Similarly, in watching Terry George’s Hotel Rwanda, I might learn about the courage and heroism of Paul Rusesabagina (played by Don Cheadle). Admiration for the real Rusesabagina is also subject to the norm. By contrast, had Adichie simply invented the suffering she depicts, or had George made up Rusesabagina’s actions, these emotions would be inapt.

23These considerations suggest a way of revising the proposal. Perhaps we can say that the cancellation context is not fiction, but imagination. So, it is when our emotions arise in the context of an imaginative project – whether in response to fiction or otherwise – that Correctness is cancelled. This would explain why pity of Anna Karenina is not subject to the norm, since we merely imagine her plight; but pity of real war victims is, because we take the descriptions to be accurate.

24The revision does not, however, rescue the proposal. The fact that an emotion arises in the context of imagining does not by itself cancel Correctness. No matter how vividly I imagine Angela Merkel happily torturing small animals, it is unfitting to genuinely despise her; similarly, imagining one’s partner to be unfaithful does not mean that anger toward them is apt (or even not inapt). Condemning someone because of actions that are figments of one’s imagination looks irrational. At the very least, cases in which we persist in feeling a certain way due to what we have merely imagined seem very much like recalcitrant emotions, provoking the same tension between emotion and judgement.

25The problem with both versions of the proposal is simple: Emotions in fictional or imaginative contexts are sometimes subject to ordinary normative constraints, but cancellation contexts are all-or-nothing; they do not make room for the selective re-application of norms. Hanks makes the point clear in further elaborating the theatre example:

As long as the play is still on there is nothing the actor can do to turn her utterances into assertions. In order to make assertions for herself the actor has to take herself out of the theatrical context by ending or exiting the play. She has to wave off the context of the play and return to the non-fictional real world (2016, 1391).

26The upshot is that the actor cannot be held responsible for what she says while in the play context. Even if she intends to make an assertion, signals this intention and so on, the utterance cannot “reach out” beyond the play; the assertive force is inevitably cancelled. By contrast, we have seen that emotions toward fiction and imagination can “escape” their contexts, in which case they are once again subject to Correctness. To accommodate this observation, we need an alternative approach.

27The invocation of cancellation contexts is consistent with a familiar distinction between the attitudes of imagining and belief in terms of quarantine (Gendler 2003). Although imagining mirrors belief in certain ways, its effects are restricted to the particular episode. For example, if Joan imagines burglars in the house, she will not take their imagined presence as further evidence of a local rise in crime, nor will she run out the back door or call the police. Whereas beliefs interact freely with other mental states and motivational systems, imaginings are confined within a circumscribed domain. Within this framework, the proposal considered in §2 amounts to treating emotions toward fiction as quarantined within an imaginative context and thereby shielded from the normative demands that govern emotions outside the context. And the problem we identified with this picture is that emotions toward fiction are not always quarantined in this way.

28The notion of quarantine suggests an isolated containment area, and it is this implication that accords with Hanks’s conception of cancellation contexts. But the border between fictional and other contexts is less like a hermetic seal, cutting off any escapes or incursions, than it is like a semipermeable membrane, selectively filtering two-way traffic. Imaginative quarantine is not absolute, as Gendler (2003) already makes clear. Still, it is helpful to avoid the implication of isolation. Borrowing a term from the psychology of narrative comprehension, we will describe emotions in response to fiction as compartmentalized within an imaginative project (Friend 2020; 2022). Compartmentalization is a matter of degree, so that some emotions may be more “contained” than others. In the next section, we rely on this idea to develop our revised proposal.

4. Compartmentalization

29Consider again the examples of emotions which we characterized as subject to Correctness: pitying real victims of the Nigerian Civil War in response to Adichie’s novel, despising Angela Merkel in response to imagining her merrily torturing animals or being angry at one’s partner after imagining their infidelity. We take it that the first emotion is fitting because people really suffered, while the latter two are not insofar as they fail to accurately represent their objects. Importantly, these judgements make sense only because the emotions are described as “spilling out” of the imaginative context, to target situations in the real world. Within the project of imagining Merkel as cruel, despising her seems perfectly apt. But refusing to vote for her on the basis of an imagined scenario would be misplaced.

30The examples motivate the following revision: Correctness is cancelled within the context of imaginative engagement with fiction so long as the emotions remain compartmentalized within the project. The context of fiction or imagination does not by itself prevent the emotions from having real-world implications, in contrast to what Hanks says about assertions. But when works of fiction invite us to imagine their contents, they also invite us to compartmentalize at least some of our emotions. For it is basic to the practice that works ask us, first and foremost, to respond with emotions directed toward the characters and events they portray. And engaging with characters and events as fiction, or as imagined, implies at least some degree of compartmentalization.

31Many works also invite us to “carry over” these emotions to real-world contexts, in the way of Adichie’s novel or George’s film (Friend 2010). But even when carryover is invited, it may not occur. Someone under the misapprehension that Adichie had made up the civil war might experience intense pity in the context of imagining the events of the narrative, without this emotion ever being directed to real-world targets. Insofar as the pity remains compartmentalized, Correctness does not apply.

32One might worry that this position removes normative constraints on any number of problematic responses, so long as we keep our emotions compartmentalized. A stark illustration comes from fictions that traffic in racist tropes. For example, Dickens’s Oliver Twist invites us to fear and despise the criminal Fagin, whose portrayal notoriously exemplifies inaccurate anti-Semitic stereotypes of Jews as essentially vile and rapacious. Similarly, Gone with the Wind (the book and film) invites us to laugh contemptuously at the slave Prissy, who is portrayed as stupid, deceptive and hysterical in line with false racist stereotypes. The two works invite us, not merely to feel a certain way about the fictional characters, but to do so because – it is presupposed – these are fitting responses to real people. Even if we reject the presuppositions and have no inclination to carry over our emotions to real-world targets, there seems to be something wrong with responding as invited. But if we say that Correctness does not apply insofar as these emotions are compartmentalized, it looks as though we cannot explain the wrongness. Our position seems to offer a worrisome get-out-of-jail free card for unethical responses.

33We agree that Oliver Twist and Gone with the Wind invite problematic emotions; however, the worry locates the problem in the wrong place. There is a distinction between what it is appropriate to do, having accepted an invitation; and which invitations it is appropriate to accept. If we go along with the fiction by despising Fagin or sneering at Prissy, the question of whether these emotions are subject to Correctness turns on the degree of compartmentalization. It is a different question, subject to different normative considerations, whether to go along at all. Let us unpack these claims in more detail.

34Like Half of a Yellow Sun and Hotel Rwanda, Oliver Twist and Gone with the Wind invite us to carry over our emotions to real-world targets. In this sense, they invite uncompartmentalized emotions. If readers or audiences do carry over their feelings, they are subject to Correctness; and in the case of the two racist works, these emotions would be unfitting. But as we have seen, they may instead “bracket” their feelings, keeping the emotions compartmentalized. Readers of Oliver Twist, for example, may accept the invitation to despise Fagin while rejecting the presupposed anti-Semitic stereotypes – that is, responding to his despicable character without taking this character to reflect a putative essence of Jewry. Someone unaware of the stereotypes would presumably accomplish this with no difficulty. Insofar as the emotion remains compartmentalized, it is not subject to Correctness.

37These concerns do not entail that compartmentalization is never appropriate, even in response to ethically problematic fictions. It is an unfortunate fact that such works abound across the history of novels, plays, films and so on. Insofar as we want to appreciate such works on their own terms – that is, to enter imaginatively into their worlds, even if we also adopt a critical stance toward this engagement – we have reason to compartmentalize our emotions.

38It is outside the scope of the present discussion to say what we ought to do in these cases, all things considered. Our point is just that the question of whether compartmentalization is appropriate in any given case is distinct from the question of whether emotions, having been compartmentalized, are subject to Correctness.

39To summarize: According to our proposal, Correctness is not automatically cancelled whenever we are responding to fiction or engaged in an imaginative project; it is not the context as such that cancels the norm. Instead, it is the degree of compartmentalization that dictates the extent to which the emotions are subject to Correctness. Fiction-directed emotions evade the normative tension that characterizes recalcitrance simply because fiction and imagination are contexts in which we are typically invited to compartmentalize our emotions to at least some extent. In the next section, we explain why this is not true in standard cases of recalcitrance.

5. Recalcitrance again

40In the most straightforward cases of recalcitrance, our emotions are uncompartmentalized and so beholden to their usual norms. The evidence for their uncompartmentalization can be found in our actions and action tendencies. One hesitates to step on the bridge, opts for a bus ticket rather than a flight or finds oneself snapping at one’s partner after the vivid dream. As noted above, in such cases, how we take things to be and how we are reacting emotionally come into tension – either things are not as they are taken to be or the emotion is an unfitting response. This conflict stands in contrast to our engagement with fiction. We might jump at the sight of Jaws or feel the pangs of guilt or sadness when reading a novel, but when we are compartmentalizing as invited, we don’t hesitate to go for a swim or send letters of condolence to fictional characters. Compartmentalization within an imaginative project can suspend the ordinary norm of Correctness for the contained emotion, thereby dissolving the pairwise tension.

42These observations do not entail that cases of recalcitrance and emotions towards fictions are simply on a spectrum. An important difference lies in the implicit invitations to compartmentalize as well as our acceptance of those invitations. Fictional contexts typically invite compartmentalization as part of the accepted mode of engagement. We are encouraged to enter an imaginative world, and the emotions experienced therein are shaped by the content of that world. In instances of recalcitrant emotions, however, the real-world context offers no such invitation. The harmless spider is a feature of the actual environment, not a character in a story. The partner, subject to dream-induced anger, is a real person with whom we share a real life and who, let us suppose, has not done what we dreamt they did. Not only are we not invited to compartmentalize; we ought not compartmentalize, and the usual norms of Correctness should stay in place. The normative status of the act of compartmentalizing itself is different between the typical cases of recalcitrance and fiction. We saw this distinction above when looking at Dickens’s Fagin. With fiction, there is a default invitation to compartmentalize our emotions to at least some degree. In the usual cases of recalcitrance, there is no such default and compartmentalization is not an appropriate option.

43This can be made more vivid: Suppose Jeanne confronts a dog and feels fear. The dog is not only harmless; Jeanne believes it is harmless. But now, let’s suppose, Jeanne engages with the situation just as she would if she were reading about the situation or playing a realistic virtual reality game – in effect, she pretends that this is not a real situation and compartmentalizes her fear. Our view has the consequence that the emotion no longer flouts Correctness, given that it has been compartmentalized; and we endorse this consequence. But this does not mean that Jeanne is off the normative hook. This is because compartmentalizing in this case looks entirely inappropriate. The situation does not invite immersion in a fiction scenario. Jeanne ought instead to be engaging with the real world as the real world. So, while Correctness is lifted within the self-imposed fiction, the act of imposing that fictional frame onto reality is itself normatively suspect.

44So where does this leave things? In our view, recalcitrant cases are ones in which a subject is such that either the judgement is false or the emotion flouts Correctness. Emotions towards fictions do not fit this profile because, in light of compartmentalizing, the norm of Correctness comes to be suspended and so the pairwise conflict doesn’t arise. But from this it does not follow that we can get ourselves normatively off the hook in any case of recalcitrance simply by compartmentalizing. Some situations invite and call for compartmentalization, and some do not. The permissibility and function of compartmentalization are highly context-dependent. Fiction generally invites and endorses it. Real situations demanding accurate assessment and response typically do not – at least not in the sense of treating reality as if it were fiction.

45Returning to our opening question: Why is there a normative difference between recalcitrant and fiction-directed emotions? Fiction typically invites at least some degree of compartmentalization, and to the extent that our emotions are successfully compartmentalized, Correctness is suspended. As a result, emotional reactions such as pitying Anna or feeling joy over Lincoln’s masterful slaying are permissible. In ordinary, world-directed contexts, we are not invited to compartmentalize and to the extent that we proceed as we ought to with respect to compartmentalization, we are on the normative hook for both our emotions and our judgements. Since they can’t both get things right, we are bound to flout a norm.