6 Emoting for Fictions

Abstract: The fact that we emote for fictions is incompatible with some cognitivist views of emotions and has therefore received a great deal of attention and been called the paradox of emotion and fiction. I show how the affect program theory offers an explanation of how and why we emote for fictions. Central to this approach, is the idea that the entertainment of content is prior to and more basic than the consideration of whether a content is warranted.

Keywords: content, emotion, fiction, paradox of emotion and fiction, warrant

Craig Delancey

In the last chapter I introduced a theory of the structure of the intentionality of basic emotions. I argued that some of them are fundamentally concretum-directed states, but that they can also be propositional attitudes. In this chapter, I will consider issues at the intersection of aesthetics and the philosophy of mind, questions concerned with our ability to emote for events and concreta that are depicted in fictions (I will use the term "fiction" to refer to any work which is understood to refer to unreal events; hence, I include stage dramas, novels, and films). I hope to accomplish three tasks. First, this is a problem that will help flesh out some of the details of heterogenous intentionality. Second, the heterogenous intentionality of basic emotions can help explain two outstanding problems at this intersection, which I call the problem of reports, and the problem of narrative demand. Third, our ability to emote for fictions is inconsistent with cognitivism about emotions, and so reviewing these issues provides another powerful argument against cognitivism and related views of mind.

There are other things to be observed in our exploration of this issue. A naturalist view of emotions like the affect program theory suffers from the prejudice that it will fail to adequately account for what matters in our emotional lives. How could a theory of emotions that claims that the relevant emotions are pancultural, inheritable, and fundamentally biological phenomena, shared by other kinds of animals, do justice to our experience of emotions, or their important role in a moral life, or to the aesthetic aspects of life? I showed in chapter 3 that one influential irrealist theory of mind, interpretationism, fails to explain quite unremarkable emotional behavior. In this chapter, I can show that reductive or doxastic cognitivism fail to adequately explain—often fails to even allow for the possibility of—something so human as our ability to emote for fictions.

Many of the issues that I raise in this chapter shall ultimately be decided by empirical research. However, there is much that can be clarified by a philosophical analysis, given what we already know. Also, some very plausible hypotheses can be made to explain some aspects of emoting for fictions, and these hypotheses are suggestively consistent with some of the

end p.103

 

other views developed in this book. Thus, in this chapter I stray a bit from what is known, to indulge in some (what I hope are useful and plausible) hypotheses concerning these abilities.

Emoting for Fictions and Cognitivism

I showed a number of objections in chapter 2 to doxastic cognitivism and to reductive cognitivism, some relying upon scientific results. But there are things commonly observed and experienced by all of us that entail very obvious objections to these theories. One of the most interesting of these is music. Music need not be about anything to generate affects; we hear a particular melody and, as if entrained to the dynamic of the flow, our affective state can change. There is no plausible way for a reductive or doxastic cognitive theory to cohere with this possibility: no beliefs are required, it is unclear in what sense any complex cognitions, such as other propositional attitudes, could be required. If music can generate emotions, then these kinds of cognitive theories are simply false.³² This alone is a compelling mystery that is, more often than not, ignored by cognitivists. But in this chapter I will be concerned with another anomaly: that we emote for the concreta or the events portrayed in fictions. If a basic emotion were to require, or were in part even constituted by, a particular kind of belief, then how can it be that we frequently have that emotion for situations in which we explicitly understand that the relevant proposition is false? Typically, for doxastic cognitivism and related kinds of reductive cognitivism, for someone to be angry, she must believe that someone has been wronged; to feel sad, she must believe that someone has suffered some loss; to feel fear, she must believe that something of value is in danger. The problem for emoting for fictions, therefore, is that we have (what some will say only appear to be) emotions for fictional characters and situations. If Karen is angry that King Lear's daughters are cruel to him, and Karen is sad for the king, then on the reductive or doxastic cognitivist theory of emotions she must believe that King Lear's daughters are being cruel to the king and that the king is suffering as a result. But Karen knows that King Lear is just a play, that she is sitting in the theater, that John Gielgud is an actor and not a king, that he does not suffer (at least not in any direct, morally culpable way) the harms being portrayed. Presumably she also does not believe that King Lear was an actual person. So what exactly is happening when she frowns, and turns red, or even when tears come to her eyes? As Bijoy Boruah has put it: "Why is it that the sadness, which is defined in real-life contexts by reference to an appropriate belief about the sad object, recurs in a context that excludes the rational possibility of forming the appropriate belief?" (1988, 83). On pain of contradiction, some presupposition must be rejected—either the claim that Karen is having an emotion, or the claim that Karen does not believe that King Lear exists, or the assumption that the appropriate kind of belief is necessary for the emotion.

end p.104

 

Attempts to Save Cognitivism

A host of solutions to this puzzle have been proposed. The majority of these solutions struggle to retain a cognitive theory of emotion intact. Here are five of the leading offerings:

1. We are irrational. Colin Radford, whose 1975 paper "How Can We Be Moved by the Fate of Anna Karenina?" may be the origin of the contemporary debate, poses the emoting for fictions problem in the following way: Suppose a man tells us a harrowing story about his sister, and we are harrowed. But then, after completing the story and allowing it to have its effect, the man tells us that he made the whole thing up. Radford suggests that we would be relieved, and not harrowed (although we might have other emotions in addition to relief, like embarrassment or anger that we were deceived). He suggests the same is true for all these kinds of cases: if we believed P and had an appropriate kind of emotion for and because of P; then upon learning P was false we would normally stop having the emotion. These cases, Radford believes, are like our emoting for fictions in that we would feel the same kind of emotions were the fiction true and we were witnessing it or heard a report of it, but with the odd additional fact that we know the fictional situation is false, and the emotion does not end because of this. (In fact, the situations are perhaps not analogous: I return to this point below.) We might suppose that two narratives were told, both of them identical, but that in the one case we are led to believe that the narrative is a true account—let us call this kind of narrative a report—and therefore has an implicit claim to being true, and in the other we are told the narrative is a fiction. In the case of reports, Radford claims, we would find the emotion we experience defused when the relevant beliefs were defeated. (I will call the claim that we emote more strongly for believed than for disbelieved contents, and the consequent issue of explaining this, the problem of reports. I discuss it at more length below.) In the case of fictions, however, we ostensibly do not believe the content from the beginning. But if we fix the relevant background desires in the two cases,³³ it would seem that the cases could have the same contents and be thus relevantly the same. Why do we then have the emotion in the case of the fiction? Why does the disbelief not rule out, or at least impair, the emotions in the case of fictions as it supposedly does in the case of a report?

Radford implicitly argues for at least doxastic cognitivism by generalizing from examples of reports, where it may seem plausible that the defeat of the relevant belief would defuse the emotion. However, he could just as easily have rejected doxastic cognitivism by generalizing from the cases of emoting for fictions. Given that he takes the report case as primary, then he must explain what is different in the fiction case. Radford's solution is that we simply are inconsistent when emoting for fictions; he solves the puzzle of emoting for fiction by holding that we do believe in the relevant content of the fiction, and since by supposition we disbelieve it, then we are inconsistent.

end p.105

 

For example, Karen is inconsistent because she both believes that King Lear does not suffer (because she believes he does not exist) and believes that King Lear does suffer (evidenced by her emotional reaction).

But it is not clear how this "inconsistency and . . . incoherence" (1975, 78) plays itself out in Karen or any other emoter for a fiction except in that they emote for the fiction. Karen might be an impeccably rational person in all other matters, a logician by trade, continent and calm. In what sense is she prone to inconsistency and incoherence except in this case of the emoting for fictions? Inconsistency is posited only for the purpose of saving an implicit cognitivist theory of emotions. As Radford himself reminds us when drawing out the differences between emoting for fictions and for believed reports, we do not charge up onto the stage to beat (the actor playing) Goneril or warn (the actor playing) Gloucester. Thus, according to him, we are inconsistent, and act as if we believe Lear suffers in as much as we have the emotion, but we show no signs of this inconsistency aside from the emotion. We don't make false claims, we don't behave otherwise as if Lear exists, and so on.

2. The "emotion" for the fiction is not a genuine emotion. Kendall Walton has described vividly the puzzle of emoting for fictions:

Charles is watching a horror movie about a terrible green slime. He cringes in his seat as the slime oozes slowly but relentlessly over the earth destroying everything in its path. Soon a greasy head emerges from the undulating mass, and two beady eyes roll around, finally fixing on the camera. The slime, picking up speed, oozes on a new course straight toward the viewers. Charles emits a shriek and clutches desperately at his chair. Afterwards, still shaken, Charles confesses that he was "terrified" of the slime. Was he? (1978, 5)

Walton complicates the usual puzzle by supposing that Charles fears for himself. This is a very different situation from having care for King Lear or fearing for Gloucester. It is not impossible, or even improbable, that Charles might so fear for himself (although the description is a bit unrealistic in its histrionics). However, so as to keep our focus on the puzzle, I shall reconstrue the case for now to be that Charles fears for the protagonist of the film; it shall otherwise be relevantly the same.

By supposition Charles never believes the slime is real. He also has such physiological features of fear that—setting aside any physiological changes distinguishing believing that P and not believing that P—it seems that Charles's physiological state is going to be indistinguishable (along all the relevant measures) from the kind of state he would be in if he feared and believed in the slime. But Walton insists that Charles's affective state is not actually fear, but rather is "quasi fear." Quasi fear is distinguished from real fear because in the cases of quasi fear the subject is only make believing that

end p.106

 

the relevant situation is actual. Walton has a theory of make-believe behavior, but for our purposes here the relevant distinction is merely that Charles by supposition does not believe the relevant propositions that Walton supposes are required for his fear (that the slime exists, that the protagonist is in danger, etc.). But since the physiology of the state of fear and quasi fear are relevantly similar, it seems that we have here two affective states, of identical kinds except that, for the one state, belief can be ascribed to the subject, and, in the other, the relevant proposition is entertained but not believed. Of course, Charles does not flee the theater, so his behavior is not indicative of the most extreme kinds of fear. But Charles might also fear great heights, rightly know that they can kill him should he fall, and still walk along a precipice because he does not want to appear a coward. In such a case, Charles certainly is feeling real fear, even though he does not flee the precipice. Walton has not established that emotions for fictions are not like suppressed emotions or emotions which otherwise result in no action. The only difference that he uses to justify calling the one emotion a quasi emotion is the lack of belief in the content.

What is at issue for Walton seems to be taxonomy, rather than some substantive point: a way of dividing emotions into two classes, distinguished only by their relation to their contents (or, more accurately, the agent's other relations to the content). Cognitivism would be saved because it would turn out to apply only to those emotions that have believed content, and not those "quasi" emotions which do not—in other words, doxastic cognitivism would be not a theory of emotion but a way of classifying emotions. But Walton denies that this is what he is doing. "The issue is not just one of fidelity to a deeply ingrained pretheoretical conception of fear," he argues. "The perspicuity of our understanding of human nature is at stake" (1990, 202). So that if we are to assimilate different instances of affective states that are fearlike together, including states of fearing for fiction, and being afraid of a genuine danger, and so on, we will "emphasize superficial similarities at the expense of fundamental differences" (202). However, it is hard to see how we can call the differences in belief fundamental, and label "superficial" the many physiological and behavioral features that he seems to allow by presupposition are relevantly identical. The question is begged in favor of doxastic cognitivism.

Thus, Walton's claim that our emoting for fictions is mere quasi emoting is at best a taxonomic move. But there is little reason to accept this standard of two different categories of emotion. Our normal discourse does not use this distinction. What is special about emotions—what separates cognitive emotions from other propositional attitudes, for example—seems to be present in both the case of emoting for a report and emoting for a fiction; after all, Walton calls them both "emotions." More important, Walton's move does not solve the real puzzle: even if we accepted his terminology, it still remains true that we have these affective states, the "quasi emotions," for fictional contents, and we might very well be puzzled by that and wonder how it is possible.

end p.107

 

3. We emote that these kinds of things actually happen. Michael Weston, writing in reply to Colin Radford, has given a subtle account of emoting for fiction, accepting that these are genuine emotions, and rejecting that we are irrational when we so emote (1975). Weston argues that what actually happens when we emote for fiction is that we both emote for the content of the fiction inasmuch as we recognize that such things do or can happen in our own actual world, and that this recognition requires that, or is successful to the degree that, the plot of the fiction develops the interconnections which give to the relevant fictional situation its plausibility and meaning.

Weston is one of the few in this debate who has drawn attention to the very important issue that fictions for which we emote have narrative structure. For example, we feel no pity if told the following story: "King Lear was a king who suffered terribly from two ungrateful and wicked daughters." The focus upon what is false in fiction can convey the sense, however, that it is for, or because of, key propositions like this that we seem to emote. I will return to this insight below, where I will call the issue of explaining this the problem of narrative demand: that the degree of our affective reaction to a fiction is dependent upon some narrative skill.

However, although there is surely some connection between our emoting for the fiction and our understanding that some of these situations can occur in the real world, this solution is itself inadequate. For, as Robert Yanal has observed (1994, 56), I might get frightened at a vampire film, or be angry at the dragon denizen of a fantastic world in a novel, knowing very well that there are no things like vampires or dragons. Weston might well respond that something like what happens when the vampire or dragon attacks happens under other situations—such as in murder and forest fires. But this would have two flaws. It seems that it is in part the unique, essential features of these creatures that I emote for, and not bloodletting or fire in the abstract (Yanal 1994, 58). Furthermore, weakened enough in this way, the theory would start to be far too vague to be viable: pressed to the extreme by using a long list of fantastic, nonexistent objects—like vampires or dragons—we would ultimately have the object of fear being the fearful, the object of anger being the infuriating, and so on.

Finally, there are situations for which we emote which simply cannot occur.

4. We emote for a possible world or for counterfactual situations. Some scholars have suggested that the way to understand fictional discourse is as discourse about possible worlds. One might presume, therefore, that the case of emoting for fiction is quite simple: the puzzle is solved by qualifying the supposition that we don't believe in the relevant content in the fiction. Instead, we do believe it, we just believe it is true of another possible world, not our own.

This is the most easily refuted of solutions to the emoting for fictions problem. There are many fictions which describe impossible situations. Examples include Italo Calvino's "Tutto in un punto" ("All in a point")

end p.108

 

(Calvino 1993), which dramatizes the crowded conditions at the beginning of the universe, when all living beings exist in one point. The geometries described (e.g., people getting into each-other's way) are logically impossible. Certainly we can have emotional reactions for these or other impossible, or impossibly situated, characters; and it follows that in such a case we cannot be emoting for a possible world.

5. Reject cognitivism about emotions. The four proposals above attempt to save doxastic or reductive cognitivism; they appear to be motivated by the assumption that cognitivism is true. But another solution to the problem is to reject cognitivism's belief condition and therefore to reject doxastic cognitivism and the relevant kinds of reductive cognitivism. Several philosophers have suggested this approach, including Roger Scruton (1974), Peter Lamarque (1981), and Robert Yanal (1994). Here I will consider the two more recent views.

Yanal proposes what he calls "realism," which is characterized by rejecting the proposition that "we feel emotions towards characters and situations only when we believe them to be real and not fictional" (1994, 69). The puzzle for Yanal then becomes something more subtle: what distinguishes our emotions for fictions from our emotions for reports and actual observed or experienced situations? Yanal suggests that our knowledge that the fiction is a fiction does it: these emotions are incapable of consummation or proper expression. We cannot treat the fictional character, or react to the fictional event, in the way we can treat or react to actual ones. Thus:

Walton thinks pity, anger, or love towards fictions less than real, mere quasi feelings. But I think that what we have is real pity that must be kept to oneself, real anger that is forever ineffectual, real love that is never to be returned. There is a sort of pathos that often permeates emotion towards fiction, a kind of pensive nostalgia, bordering sometimes on the melancholy, sometimes on the bittersweet—exactly the sort of pathos one expects to find with passions incapable of consummation. (74)

This insight might still leave us wanting for a more elaborate account of what kind of thing the fictional object of the emotions is. Lamarque has gone some steps in this direction by suggesting that fictional propositions are like Fregean thoughts: "Fictional characters enter our world in the mundane guise of descriptions . . . and become the objects of our emotional responses as mental representations or, as I shall call them, thought-contents characterized by those descriptions" (1981, 293). Lamarque then distinguishes between emoting by, which applies to the real object of the emotion, and emoting of, which applies to the intentional content of the emotion. Being frightened by a tiger requires that the tiger exist; but being frightened of a tiger, Lamarque claims, does not require that the tiger exist. And, in the case of fictions, the intentional content becomes the real object,

end p.109

 

Gottlob Frege

Gottlob Frege (1848-1925) was a mathematician and philosopher. In addition to inventing much of modern logic, Frege offered careful analyses of concepts, reference, and meaning. His distinction between sense and reference—in more contemporary terms, between meaning and reference—is an important staple of any consideration of reference. By "thoughts," Frege means something like our notion of a proposition: a thought is the meaning of a complete sentence.

and we are frightened by the mere thought that is evoked by the fiction. Thus, he holds—as Frege did—that reference in fiction is indirect reference: it is reference to sense.

I believe that Lamarque's approach is essentially correct in that it holds that the content of the propositions of fiction, and not the usual referents, is that for which we emote in these cases. However, a general endorsement of a Fregean position will leave a number of outstanding problems. First, if we take it that fiction actually is indirect reference, then there is a formal problem with the not uncommon kind of fiction where stories are told within stories. Second, merely observing that we emote for the fictional content does not explain why the features of good storytelling are requisite for emotions to occur.

A Naturalist Explanation

We have seen that rather pedestrian examples of emotions provide strong counterexamples to doxastic cognitivism and to some forms of reductive cognitivism: we can emote for a fictional situation when we fail to have the beliefs that most forms of cognitivism entail are necessary. Claims to get around this are not credible, and reveal more an eagerness to save an erroneous theory than any plausible insight into emotion. Can the affect program theory do better?

The first thing to note is that the affect program theory does better by default because it simply does not allow for the formulation of the problem in the first place. The affect program does not require that the agent believe a proposition for it to cause an emotion. Instead, we have granted only that basic emotions can be propositional attitudes, and (what may be an equivalent statement) that they can be caused by beliefs or other propositional attitudes. Furthermore, since basic emotions have a heterogenous intentionality, they can also have as their object concreta, which need not be actual; again, emoting for fictional concreta is not a problem, since we have not assumed the concreta must be actual.

But it would be helpful to actually flesh out these claims, since it seems intuitive that we should emote for beliefs and actual objects first and foremost.

end p.110

 

Thus, in what follows, I will use the notion of the heterogenous intentionality of the basic emotions and of the hierarchical theory of mind to explain in part why we can have affects for fictional events and concreta. The case of emoting for fictions provides a way to explore more clearly the nature of the intentionality of the basic emotions.

A Neo-Fregean Account of Emotional Content

Peter Lamarque offered the fruitful suggestion that we see emotions for fictions as emotions for fictional contents. Paradigm cases where Frege's indirect reference occurs are in opaque contexts. If Karen believes that Malcolm X is a Moslem, it does not follow that Karen believes that Malcolm Little is a Moslem, since she may not know that Malcolm X is Malcolm Little. For Frege, a proper name that appears in such a context cannot have its usual referent; instead, its referent is its sense. This fits nicely with the fact that we treat, for example, "King Lear" as a name, but one without an individual King as its reference. However, this doctrine quickly leads us into trouble. For, if Tim believes that Karen believes that Malcolm X is a Moslem, then it seems by simple recursion of the opaque contexts that what "Malcolm X" refers to in this thought is the sense of its sense. Whereas there is an intuitive plausibility, largely arising from the great utility of distinguishing between sense and reference, in taking the reference of a proper name to be its sense when it occurs in such contexts, it is obscure what a sense of a sense is, or what a sense of a sense of a sense is, and so on. Michael Dummett has suggested that this problem of compiled senses of senses threatens to refute the very approach:

Since we cannot say what the simply indirect sense of an expression is, we cannot even say what its referent is when it occurs in double oratio obliqua; it would seem to follow that we cannot even know how to judge the truth-value of a sentence involving double oratio obliqua. This constitutes a reductio ad absurdum of the whole theory. (1981, 267).

If fictions operated as indirect reference in a Fregean framework, then we would have this same problem. Chaucer's Canterbury Tales is a sequence of stories told by narrators who are all characters in one overarching story. If fictional discourse is indirect discourse of the same kind as other opaque contexts, then it follows that each story within the story conveys senses of senses. Again, it is not at all clear what such a thing is, and there does not seem to be any relevant difference in our reactions to a fiction as a result of whether it is within another fiction or not.

A logical framework and accompanying cognitive theory that will escape this problem has been developed by Nino Cocchiarella (1995). Cocchiarella suggests that the best way to model fiction is to treat fictional discourse as happening within a fiction operator: (In the story S . . . ). Of course, that a

end p.111

 

narrative is a fiction is usually given by convention, so we do not have to preface the fiction with any such phrase; the operator is implicit. But one thing this operator captures in a logical representation of fictional discourse is that we do not have sense of senses piling up, but rather whatever happens within the operator, regardless of how many levels of stories within stories there are, is all sense (to use Frege's term) of the same order.³⁴ The principal import of the fiction operator is not to draw our attention to the fact that we will focus upon the story's sense alone, but rather to indicate that we understand that what happens within a story is to be taken as having deactivated reference. We understand that in the story The Hound of the Baskervilles, the term "Sherlock Holmes" does not actively refer to anyone in the actual world. However, the term does still have a sense—it is a referential concept that the reader understands. Fiction is therefore a relation of contents that lack active reference, and readers can care about and have emotional reactions to those contents.

Our reactions to fictions, inasmuch as they are reactions to propositional contents, are not therefore reactions about possible worlds, nor about the actual world, but about a human construction of intensional contents. And as a result, another advantage of Cocchiarella's approach is that we can meaningfully say or think about impossible things in the fiction. This does not cause logical confusion because we understand that the reference to such objects lacks active reference. Allowing for this is necessary for any realistic theory of human cognition, since fictions incorporating impossible objects exist, are understood by us, and therefore are meaningful.

What is it for a referential concept to be deactivated? Presumably in adults it is understood that the terms are meant to have a sense like their activated counterparts, and that the terms can play the same kind of roles in sentences as can actively referring terms, but they have no referent. Of course, someone may lack an explicit concept of reference or activated reference and yet still understand the distinction between a fiction and a report. What separates the two kinds of narratives is at least an understanding that the things referred to by deactivated referring concepts cannot be encountered in the world, that they cannot effect changes in the actual world, and that we can do nothing to make changes in or for the referent of the deactivated referring concept. Thus, we could read "deactivation" as indicating that these objects are not active in our world and that we cannot act upon them. We understand that though we might be able to help the crime victim we hear about on television, or we might be a victim of the same criminal, we cannot help von Helsing in his struggle with Dracula, nor can we ourselves affect or be affected by Dracula.

The Fiction Operator and the Ability to Recognize the Unwarranted

I argued in the last chapter that some basic emotions can have both concreta and events or states of affairs as their objects. We can now see how this

end p.112

 

facilitates the ability to have emotions for imaginary objects and situations. According to the neo-Fregean approach, we emote for the contents of the relevant referential concepts, or of the proposition, that is the intentional object of the emotion. It may seem paradoxical that we would take these inactive referents as the objects of emotions; but this would be so only if we had to, in some sense, alter the normal function of the relevant affect in order to get it to take the imaginary concretum or event as its object. That is, as the problem of our emoting for fictions is normally presented, one might have the impression that belief is a fundamental or primordial cognitive state, and that we naturally only emote for the believed contents and have to learn to emote for fictions—a kind of corrupting irrationality learned as a form of entertainment. This is particularly compelling if we assume a highly cognitive view of mind, and see autonomy and the normal function of the mind as arising from having the correct kinds of beliefs and using these to guide action.

The truth, however, is to flip this picture on its head. We can refer to both development, and homology, to establish this. Children of a very young age do not yet understand the deactivation of reference in a fiction.³⁵ It is because of this that, until a child has learned to understand the difference between fictional and actual content, a parent is well advised to protect the child from frightening fictions lest both child and parent find sleep impossible. Observing this fact, we might just say that the child believes all content. This makes some sense, but without the notion of deactivated reference, or merely entertained thoughts, the notion of belief is not clearly defined. A better explanation is that the child has the ability to entertain propositional contents, and treats them all as of the same value, and that from our perspective this treatment is quite like the kind of commitment we reserve for belief. The child must in turn learn to separate out some thoughts as warranted beliefs, others as fictions; that is, to have the ability to apply the fiction operator is essentially related to the ability to apply notions of warrant to propositions. To have beliefs and acknowledge them as warranted is a learned and special skill; and, just so, to entertain content for content's sake, while being aware that it is merely entertained because it is unwarranted, is a learned and special skill. We understand that that is what we are doing when emoting for a fiction. This is consistent with the plausible posit that nonhuman animals which both share with us homologs of basic emotions,³⁶ but lack the cognitive skills necessary for distinguishing warranted from unwarranted contents, are incapable of merely entertaining propositional contents. That is, there are no dog dramas and there can be none, because we cannot train a dog to separate mere pretending from actual activity (at least, not for enough activities).

Thus, our experience with development, and the special case of emoting for fictions, suggest the following hypothesis: The ability to distinguish belief and the related ideas of activated reference and satisfaction (that is, whether a proposition is true or false) are the more advanced cognitive skills, and merely entertaining content is the more fundamental and developmentally

end p.113

 

prior capability. Distinguishing between active and deactivated reference, and warranted beliefs and mere thoughts, is something that must be learned or at least that must develop. One must have the idea of false propositions, and then of imaginary propositions, to be able to formulate a notion of deactivated reference in the first place.

At the level of the subcognitive emotional reaction, warrant and the fiction operator probably play no role. If one has a subcognitive fear of (the actress playing) Goneril or a subcognitive anger at (the actor playing) the conspirator, the affective reaction, although perhaps weaker for various reasons (e.g., because we suppress it), is not fundamentally different than would be the same reaction for an active referent. The point is that the idea of warrant simply may not apply at that level. However, with our ability to distinguish warranted from unwarranted beliefs or imagined propositions should arise the ability to understand deactivated reference. Does this ability change our emotional reactions to such contents?

The Problem of Reports

We can now see why Radford is wrong to suggest that false reports and fictions are analogous. It is an empirical question whether, and to what degree, our emotions are more intense when they have as content a proposition that we believe, as opposed to one that we do not. I do not claim to know one way or the other. It may seem plausible that we emote more strongly for believed contents; to be told a harrowing story about a friend is quite a different thing from being told such a story about a fictional character just introduced to the reader (say, on page 1 of a novel). But the difference here may entirely depend upon the greater associations that one has in the former case. It may well be that we care a great deal more, and emote much more intensely, for well-developed characters in a beloved novel than for some person known only in name, about whom we hear on the evening news. This is a question that cries out for empirical research.

However, let us suppose that there is some difference in the intensity of our emotions for believed versus disbelieved contents. We normally understand from the beginning of a story that we are dealing with a story and hence the propositions we will hear are in the scope of the fiction operator and the referring concepts are deactivated. If there is any doubt, we may find ourselves in a strange emotional state, wondering whether the narrative is a report or a fiction, and—like people passing a man lying in the street—unsure of whether we should get involved. This does not explain, however, why we might react differently to a report that turns out to be false than we do to a fiction that we understand is a fiction from the beginning.

If we do show a differential response to believed and disbelieved contents, it is reasonable to suppose that this is a product of our learned ability to separate out warranted from unwarranted cognitive contents. Thus, we may indeed be suppressing, in some sense, our emotional reactions in those cases where the content does not warrant belief. In the hierarchical model

end p.114

 

of mind, this would entail something like our learning to suppress the connection between cognitive contents and the affect as part of learning to distinguish warranted from unwarranted contents, and active from inactive references. If there is a differential reaction to believed versus disbelieved contents, then, we can hypothesize that we emote more for a report than a fiction, or at least differently for a report than for a fiction, because we have learned to (or perhaps developed the ability to) suppress the emotion-eliciting nature of disbelieved contents. Emoting for fictions is not then an irrational aberration or a heroic feat of disbelief; but it may involve a relaxation of the suppression of the emotional import of the relevant contents. To turn the phrase: enjoying a fiction is not a willing suspension of disbelief, but rather a willing suspension of affective suppression. Thus, perhaps a withdrawn report will not evoke emotions like a fiction will because we reassert the suppression of these imports; and, given the social impropriety of lying, we should expect this reasserted suppression to be vigorous, since the liar in Radford's story (who told a false report about his sister) would have essentially insulted or at least mocked the listener. There is a vulnerability revealed that one would want to quickly eliminate; one way to do that would be to suppress the emotional reactions that one had, in part in sympathy, for this person.

The Problem of Narrative Demand

These tools, and the heterogenous intentionality of the basic emotions, allow an explanation of the problem of narrative demand: although we do not need to believe the propositions of a fiction, nor believe that the concreta in it are actual, we do require that the fiction be developed in the correct way. To repeat my previous example: we emote for the characters of Shakespeare's King Lear and not for characters in a proposition like "King Lear suffered at the hands of his ungrateful daughters" or "Gloucester had his eyes stamped out by evil conspirators." Evoking emotions requires that some narrative demands be satisfied: that a good story be told and told well. Understanding this requires special attention to the heterogenous nature of emotional intentionality. Inasmuch as art can appeal directly to the subcognitive emotions, the basic emotions can be elicited more directly, although the ability to exploit these subcognitive elicitations itself creates a kind of narrative demand. Inasmuch as art appeals to the cognitive elicitation of basic emotions, it must appeal either to subcognitive elicitation indirectly through cognitive elements, or it can portray prototypical events or states of affairs that typically cause the emotion (given that conditions are appropriate).

In focusing upon written fictions, such as novels, we might easily pass over several other features of how works of art can generate emotions in us. One way, perhaps best exemplified in film, is through stimuli that do not appear to need to generate any relevant kind of conscious cognitive content in order to be effective. There appear to be several distinct ways that this

end p.115

 

may occur. First, stimuli may either directly prime the affect, or be excitatory of the relevant affect. Thus, a frightening film can take advantage of dark scenes, sudden loud noises and motions, and unsettling sounds and music. All of these things appear to either prime or elicit fear reactions in a way quite independent of whether the viewer entertains the relevant kinds of propositions; that is, it does not appear that the viewer need conclude, "a dark place is frightening," "a loud noise is frightening," and so on. If this is priming, then the subject is being put in a state in which he is more disposed to have the relevant emotion; if it is directly excitatory, then the subject is actually having the emotion as a result of these stimuli. Both are possible; future empirical research may settle the issue. In either case, the artists are knowingly undercutting our cognitive filters (and we encourage them to do so), to evoke basic emotions and other affects in direct ways that are difficult or impossible for us to control.³⁷ Second, the appropriate object of our emotion—such as a frightening person—can be presented directly, and in a way that fixates our fear upon the object without actually appealing to any propositional constructions. The antagonist may be a man who sneers menacingly at all times, and is very strong. In such cases, we are given a concretum for the emotion without actually any direct appeal being made to an inferential chain concluding with, for example, the proposition that this concretum is a threat. We can think of this as a way in which the artist can exploit the subcognitive aspect of concretum-directed intentionality.

For written fictions, these kind of avenues are not available: everything must pass through cognition, being taken in as sentences, or at least as language. But the successful novelist understands that there is nothing frightening in saying the protagonist is in danger. Rather, the novelist has other tools at her disposal. One is to build ambience in a way somewhat similar to the way the filmmaker does, but which clearly relies upon cognitive capabilities. A novelist can build a frightening ambience, for example, through careful details. How this fosters emotions demands further study. However, one speculation is tempting: perhaps the novelist evokes cognitive contents (like those which the filmmaker actually shows) which, even if not directly ostensibly frightening (that is, the reader may not believe that a dark night is frightening), are associated with frightening contents. That is, the subcognitive kind of priming of the relevant emotions that can occur in films can also be indirectly stimulated by cognitive contents that refer to the same (kinds of) stimuli. We can hypothesize that these emotive contents are exploited by stimulating a semantic network of connections which can allow for increased stimulation of the contents associated with the relevant emotion, and ultimately for the elicitation of that emotion.³⁸

More important, in all these artworks, the artist has at her disposal the recreation of the very kinds of situations which if actual would evoke emotions; what psychologists sometimes call prototypical situations. Thus, for example, a prototypical scene that may evoke anger would be if someone

end p.116

 

whom we value is mistreated. The novelist carefully elicits these values by both getting us to sympathize with the protagonists, and thus to value them and their motives, and by then placing those protagonists in carefully constructed prototypical scenes. In the fiction, then, the characters are treated to situations that are quite like reports, with the exception that the reference is not active. Supposing that active reference is unnecessary will give us the useful result that an account of the cognitive elicitation of emotions in both situations can be essentially the same.

The one difference, of course, is that we need the additional posit that the basic emotions can be had in sympathy. This is something which appears to me an extremely plausible hypothesis, since it is, as far as I can see, entailed by the fact that we do emote for the protagonists of fictions. Also, this hypothesis has some very important corollary explanatory power. For example, it suggests another possible explanation of the role of cognitive emotions: that increased representational complexity serving basic emotions may not only have the benefit of allowing an extension of the utility of those emotions (as suggested in the last chapter), but also may have evolved with our ability to understand other minds and to organize social bonds with them. Thus, some instances of basic emotions with cognitive content may function foremost through a sympathetic modeling of the concerns and values of others, and thus evoke sympathetic emotions in us. Also, understanding sympathetic emotions can be of fundamental importance to understanding important aspects of moral motivation (I will return to this possibility in chapter 8).

Conclusion: Weak Content Cognitivism

In chapter 2, I introduced the term weak cognitivism as a label for the view that basic emotions can be, and perhaps often are, propositional attitudes. What these insights reveal is that weak cognitivism should be weakened even further. It is not just that the basic emotions are sometimes elicited by beliefs or are propositional attitudes that share their content with beliefs. More important, the basic emotions are sometimes elicited through the entertaining of cognitive contents, or are propositional attitudes that have as their content merely entertained propositions. And, consistent with the heterogenous intentionality of the basic emotions, sometimes we emote directly for concreta that may or may not exist. A proper theory of the relation of the basic emotions to cognition should thus be a form of weak content cognitivism: the view that basic emotions often have or are caused by propositional content, but that content need not be believed.

My exploration of emoting for fictions required several hypotheses about the mind. First, I have argued that the best explanation of emoting for fictions is that we emote for contents just like the kind of contents that would cause emotions in a report, except that we understand that the reference is deactivated. Second, the differences between a withdrawn report and a

end p.117