Disjunctivism and the Paradox of Tragedy

Abstract

In a well-known passage Sir Philip Sidney gave an ancient illustration of the strange power of the genre of tragedy:

But how much [tragedy] can move, Plutarch yieldeth a notable testimony of the abominable tyrant Alexander Pheraeus, from whose eyes a tragedy well made and represented drew abundance of tears, who without all pity had murdered infinite numbers, and some of his own blood; so as he that was not ashamed to make matters for tragedies yet could not resist the sweet violence of a tragedy.¹

Plutarch relates how Alexander left the theatre abruptly during a performance of Euripides’ Trojan Women, and ‘sent a message to the actor bidding him be of good courage and not put forth any less effort because of his departure, for it was not out of contempt for his acting that he had gone away, but because he was ashamed to have the citizens see him, who had never taken pity on any man that he had murdered, weeping over the sorrows of Hecuba and Andromache’.² Plutarch's tyrant is a useful figure to bear in mind when we attempt to answer that old problem, why tragedy gives pleasure, the so-called ‘paradox of tragedy’. In the Philebus Plato described tragic audiences as feeling pleasure at the same time as they grieve (48a5–6), Aristotle took it for granted in the Poetics that pleasure is the end of tragedy (1451b23,1453a36, b10–11), and in ancient literature generally we find ‘an awareness of the paradox that pain, as recorded in art, can give pleasure’.³

The paradox arose for Aristotle because, although in the Poetics he located the source of tragic pleasure in the emotions of pity and fear which the plot excites by means of imitation (1453b10–13), in other works he tells us that pity and fear are painful emotions.⁴ Alternatively, the paradox is set by the fact that we get pleasure from looking at representations of events that we would not wish to see in real life (1448b10–11). The connection between the two Aristotelian forms of the paradox is this: in the theatre we experience emotions and sensations which are in some interesting way related to the emotional and physical sensations that we would have in correspondent real-life scenarios, but in the theatre these experiences are enjoyable, whereas their congeners in real life would not be. Hume's way of putting the problem was this: ‘It seems an unaccountable pleasure, which the spectators of a well-written tragedy receive from sorrow, terror, anxiety, and other passions, that are in themselves disagreeable’.⁵ How to understand this phenomenon? I shall argue that we should give a disjunctivist solution to the paradox, in the sense, or one of the senses, of ‘disjunctivist’ which that word has in contemporary epistemological discussion.⁶ First, then, I need to spell out what I take the doctrine of disjunctivism, in its most promising form, to be.⁷

In the epistemology of perception a distinction is drawn between conjunctivism and disjunctivism.⁸ As I shall understand this distinction, conjunctivists think that in perception a subject is confronted, in the first instance, with an interface between mind and world: the reach of the mind extends no further than this interface, where images—for the sake of a label we can call them ‘percepts’—are located that in some sense represent or go proxy for worldly objects, the objects we ordinarily say we ‘perceive’. All being well, percepts enable the subject to acquire knowledge of the ‘external’ world (as on this approach we are forced to think of the ordinary world), but they also render that knowledge indirect: in any sense in which you can correctly be said to see a physical object in the ‘external’ world, the conjunctivist maintains that you do so mediately, via immediate cognition of quite other sorts of entity, namely purely mental percepts.

The conjunctivist, then, holds that in a successful perceptual episode there are, in addition to directly cognized representing objects (percepts), indirectly cognized represented objects (the familiar material furniture of our lives); in an unsuccessful perceptual episode, by contrast, there may be no such represented objects, or they may not be the way the episode represents them to be. A hallucination, for example, as this notion is standardly deployed by philosophers, evinces the absence of a worldly correlate in respect of at least one of the proxy entities that, according to the conjunctivist, are located on the interface in the subject's mind. What differentiates the successful from the unsuccessful case is the causal aetiology of the percepts that the subject directly cognizes: in the successful case, a percept (as) of a cat sitting on a mat, say, is caused (in an appropriate way) by a real cat in the ‘external’ world, really sitting on a real mat; in a corresponding hallucination there is, perhaps, no cat really there on the mat (or no mat, etc.).

The position earns the label ‘conjunctivism’ from the way in which it describes the successful perceptual case: that case is composed of a representation of a cat, say, internal to mind, and in addition a real cat in the world, external to mind and causing the representation that is the immediate object of the subject's cognition. The combination of mental and extra-mental elements here is a true conjunction in the sense that mental representation and worldly cause are purely contingently related. In particular, there is nothing in the conjunctivist's position that requires the cat-representation to be caused by what in fact causes it, or indeed by anything; the representation would be numerically the same entity whatever its cause, or even if it were uncaused.

Accordingly, conjunctivism is so called because it casts a successful perceptual episode as a conjunction of two factors, inner and outer. The opposing, disjunctivist position is so called not because it treats disjunctively either the successful perceptual episode in se or the unsuccessful episode in se, but because it construes both possibilities taken together, success and failure, as related disjunctively to one another. A perceptual episode is either a successful perceiving of a cat, for example, or it is in some way a failure; to keep matters simple we shall again suppose that the contrasting failure is of a hallucinatory sort, in which the subject hallucinates a cat, say. Whereas for the conjunctivist the mind at most contains a cat-representation, in both successful and unsuccessful perceptual cases, the disjunctivist holds that in a successful perception of a cat the cat itself is present in the perceiver's mind; if there had been no cat, or a numerically different cat (no matter how qualitatively similar), the perceiver's mind would necessarily have been differently configured. Mind embraces world; the ‘outer’ penetrates the ‘inner’. In the successful case, the subject's mind, at its most fundamental level of description, is already configured as a seeing of a particular, material cat, one that both the subject and we can single out and refer to as that cat; the subject's fundamental mental configuration could not have been (a contributor to) a perception of something else, or of nothing.

If a subject hallucinates a cat, the disjunctivist thinks of this subject's mind as containing a mere appearance as of a cat—only as of a cat because, in the absence of a cat, nothing can be picked out in the demonstrative or de re mode and referred to, either by the subject or by us, as that cat. On the successful or ‘good’ disjunct, as we may call it, the mind contains, and the subject is in immediate cognitive contact with, ordinary objects of perception like cats and mats; on the unsuccessful or ‘bad’ disjunct, we can think of the deluded subject's mind as containing—to the extent that the perceptual episode is unsuccessful—mere appearances. Further, a mental state like a seeing (or a purported seeing) of a cat sitting on a mat essentially has propositional content: it is essentially a seeing (or a purported seeing) that such and such is the case.⁹ So if one is located on the good disjunct one is in possession of a warrant, a justification for beliefs about the layout of the world, that one lacks if one is hallucinating. The act of my seeing a cat, for instance, provides conclusive justification for a belief to the effect that there is a cat over there, say, at such-and-such a distance and orientation from me.

Though, according to disjunctivism, a perception and a corresponding hallucination differ as a matter of what is in the subject's mind, it is conceded that there is a sense in which they may be indistinguishable to the subject; or perhaps they are simply not distinguished. We have to be careful how we express this sense: the disjunctivist should not allow that a genuine perception and a simulacrum are (or can be) indistinguishable ‘from the inside’ simpliciter. After all, for the disjunctivist the ‘outer’ penetrates the ‘inner’: the difference between the two mental states in question is as ‘inner’ a matter as can be. Nevertheless, the subject may in some suitable sense be unable to tell—or may just not tell—which state he or she is in, and provided the phrase ‘indistinguishability from the inside of genuine perception from hallucination’ is used in that sense, the disjunctivist can accept it as expressing a real possibility. Indistinguishability or failure to be distinguished ‘from the inside’ was the point of departure for the old ‘argument from illusion’; but it is now generally agreed that this argument rested on a fallacy.¹⁰ As the disjunctivist will insist, from the fact that on some particular occasion you cannot or do not tell whether you are really seeing a cat or hallucinating one, it does not follow that a strictly accurate description of the contents of your mind must be given in terms sufficiently general to cover either eventuality. All that follows is that you may not know the exact nature of a given mental state of yours, whether it is a seeing or—quite different—a hallucinating.

The point of the doctrine of disjunctivism is to close the gap between mind and world, or rather to prevent any gap from opening up in the first place. On the good disjunct there is simply no gap: mind contains world. On the bad disjunct there is, in a sense, a gap between mind and world, but it is merely a version of the gap between truth and falsity. The hallucinator's mind contains an appearance with content, the content being that of a purportedly successful perception. But at one or more points where a corresponding successful perception has de re content—it is a perception of that cat, as it might be—the hallucinator is forced, in the absence of the relevant res, to substitute purely general content—the subject hallucinates a cat, as it might be. Still, the hallucination is a would-be perception; it is an attempt to reach out and embrace the world. And although it is a failure in the sense that there is no worldly object in the place whether the hallucinator thinks there is one, the episode is a success at least to the extent of achieving content: it is, say, a hallucination of a cat, and in achieving that much the perceiver's mind is still, in a sense, reaching out and embracing the world. A purported perception is a purported perception, with all that that entails. Let me expand on this point.

Hallucination is a bona fide state of mind: it is a failure of a sort, but it is not mere nothingness; indeed it can only be a failure if it is not mere nothingness. Hallucination is to some extent an achievement: someone who is hallucinating is in some sort of mental state. But that in turn entails that certain cognitive standards are met: not just anything goes. A hallucination, after all, is defined to have content, indeed possibly quite detailed content, and certainly content of propositional form. If it is to be a genuine hallucination, its content must purport to relate de re to objects in the subject's immediate environment, objects that the subject can purportedly—though, in the case of hallucinated objects, not really—pick out using demonstrative expressions. For example, a hallucination might have the content that there is a man dressed in a particular way, perhaps sporting a flamboyant riding coat, standing up there on the battlements of that castle, with those clouds scudding by. All that content does not come for free; the subject must earn the right to dispose of it. How?

Suppose that you claim to have a perception of a man as described, but that we judge you to be hallucinating him. You are permitted to have this hallucination only if most of the time, and in most contexts, your senses are in good working order, and the hallucination is just a minor aberration. Moreover, even in the hallucinatory experience itself you really do see the battlements and the clouds, along with the rest of the environment that ‘frames’ the hallucinated man. Some such normal surround must be in place if the supposition that the subject hallucinates an object is to be coherent. Were mistakes such as the one you make when you think you see a man standing on the battlements to become much commoner in your mental life, a catastrophe point would be reached at which you would forfeit any entitlement to mental content at all. Beyond that point—in practice, of course, its location would be vague—things would just be too haywire for you even to merit the status of undergoing a hallucination. Hallucination is an achievement, as we have said, at the very least because it has definite content which can be spelled out in a that-clause; if one's mental life becomes sufficiently erratic, one can no longer claim entitlement to that content. This is an application of the Davidsonian—originally Wittgensteinian—point that mistakes, error, falsity in general, depend for their existence on their having an exceptional status.¹¹ The more mistaken someone's purported mental states are, the less inclined we will be to assign definite content to those states, and so, correspondingly, the less we will be inclined to call them mental states in the first place. A ‘mistake’ about everything, or even about very many things, would not be a mistake; as failure gets worse and worse, as it spreads its tentacles further and further, it gradually ceases to be failure, not because it becomes success, but because it ceases to be the kind of thing that could be either success or failure.

Now the disjunctivist accepts that good and bad disjuncts have commonalities. For example, as between your seeing a cat sitting on a mat and your hallucinating a state of affairs describable in those terms, its seeming to you as though there were a cat sitting on a mat—this formulation being taken neutrally, as implying neither that there is a cat on the mat nor that there is not—is a common element. But for the disjunctivist the point is that, while it is true that the two disjuncts of our scenario share a common feature, it is not the case that the perceiving subject, if located on the ‘good’ disjunct, starts, epistemically speaking, from that commonality. In the good case you start from seeing (that) a cat (is) sitting on a mat, indeed from seeing (that) that cat (is) sitting on that mat. In the successful case, your mind actually contains the cat and the mat and not mere proxies for them. We observers of this successful perceptual episode may then abstract from its detail and say that in virtue of your seeing that that cat is sitting on that mat it is a fortiori the case that it seems to you as though there were a cat sitting on that mat; we then find that this abstracted state of mind is common to the successful perceptual episode and a suitably constituted and convincing cat-hallucination.

Hence for the disjunctivist the element that is common to a good disjunct and its corresponding bad disjunct is not a ‘highest common factor’ in an epistemologically loaded sense: it is not an independent staging post that the subject has to pass through en route to acquiring perceptual knowledge. In order to count as seeing a cat sitting on a mat, say, it is not that you first, in a logical sense, have to enjoy a percept with a content that is neutral as between good and bad disjuncts—this percept being all your mind strictly contains—and that then, assuming that you satisfy some additional (causal) condition, extraneous to your mind, you are permitted to count as (indirectly) perceiving the cat on the mat. Rather, according to the disjunctivist, if the relevant perceptual episode is a success, it is in virtue of your (for example) seeing (that) that cat sitting (is) on that mat, and in virtue of your mind's containing both cat and mat (and the abstract relation sitting on), that, in the first instance, you enjoy an experience with that very content. In the second instance, by abstraction from first-instance content, you see that a cat is sitting on that mat and that that cat is sitting on a mat. These abstracted contents would also in principle be available to subjects who, respectively, saw the mat but hallucinated a cat thereupon, or who saw the cat but hallucinated a mat thereunder, provided that they satisfied the (quite stringent) conditions for being entitled to have contentful hallucinations in the first place. Abstracting still further, we can say that you enjoy an experience with the content that a cat is sitting on a mat, which content would (with the same proviso) also be available to someone hallucinating both cat and mat. (In an extreme case, we may allow that a brain in a vat has access to contentful hallucinations, that entitlement being supplied not by coeval context, but by the fact that, until recently, this brain was ensconced in a human body as part of a functioning subject and agent in the ordinary world: the entitlement to content that was then earned by this subject may be retained, at least for a short period, by the subject in its envatted existence.)¹²

Although a genuine perception and a corresponding hallucination have a common element, the disjunctivist, we have said, rejects the constructivist's conception of this common element as an intermediate stage that the subject has to traverse en route to acquiring perceptual knowledge. Perceiving does of course entail seeming to perceive, so that possession of the common element (seeming to see a cat, as it might be) is a necessary condition of perceptual success (seeing a cat), but that trivial logical relation is accompanied by a much more significant epistemological dependence running in the opposite direction: in successful perception, you seem to perceive an F by virtue of perceiving an (that) F; and your way of knowing that you seem to perceive an F goes via your knowledge that you perceive an (that) F. Perceiving, not seeming to perceive, is the successful perceiver's epistemic starting point. This reflects an important metaphysical asymmetry: the content of a bad disjunct is characterized—and can only be characterized—in terms set by the corresponding good disjunct.¹³ We might express this asymmetry by saying that hallucinating—merely seeming to perceive—involves undergoing a certain kind of privation, namely a privation of what the corresponding genuine perceiver possesses.¹⁴ Putting the point in these Aristotelian terms is helpful, I think, because possession is plainly the prior concept, and privation derivative. There are two respects in which the good disjunct enjoys metaphysical priority.

First, a subject located on a bad disjunct disposes of a mere appearance, whereas being located on a good disjunct puts one in immediate contact with reality: but ‘reality is prior, in the order of understanding, to appearance’¹⁵; ‘“seems”-talk presupposes “is”-talk’.¹⁶ The concept of an appearance presupposes the concept of what that appearance is—or, at least, of what it purports to be—an appearance of, namely a stretch of reality. Even if my ‘experience’ does not in fact reveal the world, even if it is entirely hallucinatory, still it purports to reveal the world; and it is dependent for its content (insofar as we allow it to have content) on the possibility that it reveals the world, on the possibility that it is not hallucinatory.

Secondly, and relatedly, a hallucination is, as we have said, a would-be perception. In general, the bad disjunct is essentially a failed attempt to have the content of the corresponding good disjunct in all its de re richness.¹⁷ The dependence of bad disjunct on good is then simply a reflection of the relation between success and failure: success is the prior notion; failure is defined in its terms. Success is achieving an ideal; the corresponding failure is a falling away from that ideal. In the case of a hallucination, what is crucial is that the subject is aiming to enjoy perceptual states with full de re content, but failing to achieve that content at one or more points. A perceptual episode cannot enjoy the content that is common to the successful and unsuccessful case unless it purports to be the successful case: you cannot seem to see a cat on a mat unless your mental state purports to present you with a cat and a mat that can be referred to, by you and everyone else, in the de re mode. But success cannot be constructed out of failure and some; possession is not privation and some. Likewise, the content enjoyed by a subject who is located on the good disjunct is not constructed out of the content of the bad disjunct and some; rather, the content of the good disjunct comes first, metaphysically speaking, and that of the bad disjunct is derivative. If we tried to do things the other way round, to build up content in a conjunctivistic way, the ‘appearances’ with which we started would not have content—and so would not actually be appearances.¹⁸

Note that, if we conceive the relation between good and bad disjuncts in these terms, we do not have to deny that hallucinations have phenomenal character. Some disjunctivists fear that if hallucinations are allowed to have phenomenal character, a ‘screening off’ problem will arise, which is essentially the worry that we will be forced back into a highest-common-factor conception of experience after all. But the anxiety is needless. So long as we conceive of the phenomenal character that a hallucination shares with its successful congener (an appearance, as it might be, that a cat is sitting on a mat) as being an abstraction from and so derivative of the essential content of the good disjunct (seeing that that cat is sitting on that mat), and as purporting to earn that content—so long, in other words, as we get the metaphysical priorities right—no such collapse of the disjunctivist's position will occur.

At its core, the paradox of tragedy is set by the fact that we respond with pleasure to the artistic portrayal of events which, were they real, would elicit quite other responses from us. Take the scene in King Lear in which Gloucester is blinded: if one were forced to watch any such atrocity in real life, it would be intensely harrowing; yet we go to the theatre, knowing that this blinding will be enacted on stage, and that the spectacle will in a sense be ‘harrowing’; but if the scene is well performed it gives us pleasure. I have scare-quoted the second occurrence of the word ‘harrowing’ in the last sentence because it is unclear whether the scene really is harrowing or not, and that sets us a problem. If the scene is harrowing, why do we go to the theatre to watch it, and why do we derive enjoyment from doing so? If it is not harrowing, why do we talk as though it were? The emotions of ‘sorrow, terror, anxiety’ (Hume, §1) in real life are unpleasant, but we talk as though these same emotions are felt in the tragic theatre, where they are supposedly pleasant. How can this be?

Since I shall in the sequel make use of the ‘fear’/fear example, it should be noted that in identifying this as a tragic emotion Aristotle plausibly has in mind both the ‘fear’ that we feel on behalf of the relevant protagonist(s) and the fear that we feel on our own behalf. I do not scare-quote this latter occurrence of the word ‘fear’, for what we feel on our own behalf is presumably genuine fear. It is the fear of undergoing not, perhaps, precisely the same catastrophe as befalls the tragic agent, such as killing one's father and marrying one's mother, but a catastrophe of the same general kind, involving errors, and especially cognitive errors, of a similar nature to those committed by that agent. As such—as genuine fear—this emotion is unparadoxical. The paradox attaches not to any real fear that we feel on our own behalf, but exclusively to the ‘fear’ that we feel on behalf of the tragic agent.

Many critics, from Aristotle to Hume and beyond, have assumed that any solution to the problem must be broadly conjunctivist in form: that is, they have supposed that there is an initial stage at which, in the tragic theatre, one feels (for example) real fear. According to Susan Feagin, the mechanism of tragic pleasure is this. In the theatre we experience, in the first instance, a direct response of sympathy and horror in respect of the depicted tragic events, just as we would if the events were real, but this initial response is then joined by a meta-response of satisfaction, prompted by our having (had) the direct response. The meta-response arises from our awareness of the fact that

We find ourselves to be the kind of people who respond negatively to villainy, treachery, and injustice. This discovery, or reminder, is something which, quite justly, yields satisfaction. In a way it shows what we care for, and in showing us we care for the welfare of human beings and that we deplore the immoral forces that defeat them, it reminds us of our common humanity. (2004: 188)

There are two difficulties with this account. The first is that it makes the audience's response of pleasure to viewing tragic events on stage too self-regarding.¹⁹ In witnessing the mutilation of Gloucester, one is entirely absorbed in the appalling brutality that is being enacted on stage: one's thoughts and feelings are directed outwards, not inwards. If tragic pleasure were derived from our own feelings of sympathy, then, as Hume acerbically observed in a letter to Smith, ‘An Hospital would be a more entertaining Place than a Ball’ (1932: vol. 1,313).

The second objection to Feagin's account is that it involves an implausible ‘highest common factor’ conception of the tragic response. Feagin writes, as though in reply to Hume's observation just quoted:

the meta-response of pleasure to the sympathy we feel for other people is appropriate to art but not in life because in the former there is no real suffering to continue to weigh on our feelings. In the latter case, real suffering easily commands our attention, so that any desirability of sympathy is of miniscule importance in comparison with the perniciousness of the conditions which gave rise to it. (2004: 192)

Notice the use of the verb ‘continue’ here. On Feagin's account, my initial response to tragic suffering is exactly the same as it would be if I were witnessing real suffering; that response is then embellished with a further, meta-response of pleasure (appropriate in the theatre, though not normally in real life) which has the initial response as its intentional object. As a description of the tragic experience, this account is unsatisfactory. If it were right, one would feel a momentary impulse to leap on stage and come to Gloucester's assistance. Audience members do occasionally demonstrate such naïveté: Suetonius reports a story that when Nero was acting the part of mad Hercules, ‘a young recruit who was guarding the entrance, seeing the emperor in mean attire and bound with chains, as the subject required, rushed forward to lend him aid’.²⁰ (The most famous, though possibly an apocryphal, illustration of this kind of mistake concerns a theatre company touring the Canadian Outback with a production of Othello: an audience member stood up during a performance and shot the actor playing Iago.)²¹ But in any reasonably sophisticated audience stage suffering does not even start to prompt the kind of action that real suffering does: one does not have an impulse to help Gloucester, even for a split-second, an impulse that is then countermanded by a sort of fiction-detecting superego. In real life, witnessing the pain or discomfort of another person cannot be separated, in a properly brought up person, from an impulse to assist. That impulse may of course be checked: you might, for example, judge that the ambient context was too dangerous for you to intervene. But the important point is that such situations in real life have that kind of articulated structure: the urge to help arises naturally and has to be quelled by further internal or external forces if it is not to be acted on. In the theatre, there is no such initial impulse. So the conjunctivist solution to the paradox of tragedy is unconvincing.

Let us, following Kendall Walton and others who have discussed this problem, attach the label ‘quasi-’ to the names of the emotions that one feels in response to an artistic performance as an integral part of one's aesthetic response to it.²² Now it seems clear, against Walton himself, that quasi-emotions are genuine emotions: the tragic audience is not merely pretending to feel emotion.²³ The audience that feels quasi-fear, for example, is not engaging in a pretence, or making-believe, that it feels fear (or any other emotion). But if the audience that feels quasi-fear really is feeling an emotion, is it feeling fear? Given that the audience knows (or at least believes) that the situation is not real in the relevant sense, but an imitation of reality, one would expect a negative answer to this question, and there are two interconnected reasons why that must be the right answer.²⁴

First, the experience of watching (or reading) tragic drama is deeply pleasurable. As Alex Neill puts it: ‘Any plausible account of the paradox of tragedy must involve a recognition that the pleasure and the ‘pain’ that tragedy gives rise to are in some way internally related’ (1992: 153). The experience of ‘pain’ (note the scare-quoting) and the experience of pleasure are intimately connected, in a relation of mutual dependence: there is a dependence of the pleasure on the ‘pain’ (this is tragic pleasure, after all); and, crucially, there is also a dependence of the ‘pain’ on the pleasure (it is pleasurable ‘pain’). But real pain is not pleasurable, at least not in the basic case. (The phenomenon of masochism is essentially a secondary one: here the pleasure depends on the presence of an assumed guarantee that the pain will be of relatively short duration, will not become too severe, and so on.) So the phenomenologies of real emotions and corresponding quasi-emotions are, at least in general, quite distinct. This is sometimes denied. Catherine Wilson, for example, holds that the two kinds of emotion are phenomenologically indistinguishable.²⁵ She is impressed by the fact that our experience of the real is pervaded by imaginative and perhaps even fictional elements, just as our experience of the fictional—in general our experience of artistic representations—is suffused with elements of the real. There surely is some such mutual interpenetration; nevertheless, the phenomenological difference remains. Secondly, and more importantly, quite apart from their distinct phenomenologies, an emotion and its corresponding quasi-emotion have, in general, different functional roles: fear, for example, prompts me to certain kinds of action, whereas quasi-fear prompts me, not exactly to inaction, but to very different, and much less demonstrative, types of action (continuing to sit on my seat in the auditorium, etc.). As we noted at the end of Section II, even if a successful perceptual episode and a corresponding hallucination are indistinguishable (or are simply not distinguished) ‘from the inside’, it does not follow, as the old argument from illusion assumed, that the conjunctivist therefore gives the right account of the structure of perception. Similarly, even if—what I deny—fear and quasi-fear were phenomenologically indistinguishable, that would not mean that quasi-fear was really a type of fear, or that fear was really quasi-fear and some.

So the answer to the question whether quasi-fear—the ‘fear’ that we feel in the tragic theatre on behalf of the fictional tragic agent—is a kind of fear is that it is not. But, just as a hallucination has things in common with its corresponding successful perception, so quasi-fear has things in common with real fear. A hallucination resembles the corresponding successful perception in certain respects; similarly, quasi-fear resembles real fear in certain respects. That is why we use the label ‘fear’ for what we feel in the theatre, even though in fact the emotion we feel in the theatre is fundamentally different from real fear, both phenomenologically and in its connections to action (see further Section VII). Quasi-fear is a unique, sui generis kind of emotion, pervaded by sensations of literary and aesthetic pleasure intertwined with cognitive and celebratory aspects.²⁶ Quasi-fear cannot be factored into independent feelings of real fear and pleasure that are merely externally related to one another, and which then combine to yield quasi-fear as a joint upshot. Reverting to Hume's statement of the paradox of tragedy, we may say that Hume was right that, if spectators of tragedy really did feel ‘sorrow, terror, anxiety, and other passions’—these passions being, as he says, ‘in themselves disagreeable’—then the audience's pleasure would be ‘unaccountable’. But they do not feel those passions.

In the case of fear and quasi-fear, there is no emotional highest common factor, in the relevant sense of that phrase. Of course, as in the perceptual case, we can abstract a common element between genuine fear and quasi-fear: at the very least, the disjunction of real fear and quasi-fear is itself such a commonality; any agent who is feeling either real fear or quasi-fear is a fortiori in such a disjunctive mental state. But again, as in the perceptual case, there is an underlying metaphysical asymmetry between the two disjuncts: the state of quasi-fear (and so also the common disjunctive state) is conceptually dependent on the state of real fear. This is so in ways that parallel both of the respects in which we said (Section IV) that, in the perceptual case, the good disjunct enjoyed priority over the bad. First, as we have noted, following Aristotle, art is an imitation of life: it gets its conceptual subject matter from life. (Compare: an appearance is as of reality.) Secondly, there is a sense in which art purports to be life: here I mean not the sense in which audiences sometimes naïvely mistake art for life but rather the point that art purports to present particular content; literature uses proper names, demonstratives—the whole paraphernalia of de re reference. But to the extent that the objects of discourse are fictional creations, the underlying logical form of the relevant discourse is general; for example, fictional proper names are abbreviated definite descriptions.²⁷ (Compare: a hallucination purports to have de re content; but in the places where it is a hallucination it has purely general content instead.) So quasi-emotions have a secondary, derivative status, not because they start life as their real congeners and continue to possess a real core (so Feagin), but because they depend conceptually on their real congeners.

The disjunctivist about tragic emotion rejects the conjunctivist's account of the relation between genuine and quasi-emotions, and substitutes a structure in which the agent is in one of two fundamentally different mental states, for example fear or quasi-fear. In the theatre we are presented with a stage picture which is a façade, an imitation of actions and emotions, and these imitations lack the causal background that such actions and emotions, were they ‘for real’, would have. Our knowledge (or belief) that the imitation is only an imitation, not the real thing, together with the pleasure which, as Aristotle observed, we take in imitation as such, explains why the emotion that we feel is quasi-fear, for example, as opposed to real fear. And the fact that there are resemblances between fear and quasi-fear explains why we are inclined to describe our emotions in the theatre using the language of fear, even though what we feel is not fear. These resemblances arise naturally because art is an imitation of life, which means that, were the theatrical façade genuine and a suitable real background in place, the actions we witness would elicit real fear, or rather—since what is wanted here is a normative rather than a merely descriptive connection—would be such as to render a response of real fear appropriate.

The inclination to characterize as fear an emotion which is not one of fear can be re-expressed by saying that we suffer an illusion of feeling fear. Two points are noteworthy in this connection. First, such an illusion need not conflict with our knowledge (or belief) that the situation we confront is fictional. Here, too, an analogy with the perceptual case holds: as is well known, an optical illusion such as the Müller–Lyer illusion continues to present a deceptive appearance even after we have assured ourselves that things are not as they seem, and so even though our belief conflicts with the propositional content of the appearance.²⁸ Similarly, we may continue to feel quasi-fear even as we remind ourselves that there is nothing to fear. Secondly, quasi-fear is not an imitation of real fear: stage action imitates real action, but quasi-emotions do not imitate real emotions; for one thing, quasi-fear arises spontaneously, without the necessary mental background—in particular, without formation of the necessary intentions—to enable it to imitate anything.

Thinking of the fictional case as involving the illusion of ‘real’ emotions such as fear helps us with Plutarch's tyrant (Section I). On the conjunctivist approach, Alexander would presumably start from a feeling of callous satisfaction in the sufferings of Hecuba and Andromache; that initial feeling would then be countermanded by a superego telling him that their apparent pain was just play, so that his initial response was inappropriate. He can then indulge his tears. But why should the knowledge that he is confronting a mere representation of suffering elicit tears of sympathy from the tyrant? And why should his tears be pleasurable? Not, presumably, because he thinks that his moral equipment is in good working order, as on Feagin's official line. The disjunctivist cuts free of this tangle by diagnosing the weeping Alexander as feeling not sympathy but quasi-sympathy, an illusion of sympathy. (Recall the difference from Walton's own position here: Alexander is not, contra Walton, under an illusion that he is feeling emotion.) Real sympathy would be disconcerting, not pleasurable, for the tyrant, just as real fear would be far from pleasurable for any normal person; but an illusion of any emotion can be pleasurable, if it is accompanied by knowledge of the illusion and by the pleasure that we naturally take in imitation. Of course, correlatively with the fact that feeling real sympathy would discomfit him, Alexander will worry that his emotional display could be misunderstood by the citizens: they may think he has gone soft; hence his message to the actor. A. D. Nuttall reports Craig Raine's answer to the question why tragedy gives pleasure as ‘All emotion is pleasurable’ (1996: 104); historically, this has been a common response to the paradox of tragedy.²⁹ The answer cannot be right as it stands, since many emotions are far from pleasurable, but perhaps if it is tweaked to say that all (or much) quasi-emotion is pleasurable, then it is more plausible.

Another advantage of using the language of illusion in the context of the paradox of tragedy is that it marks an important similarity between the two kinds of scenario that I have been subjecting to disjunctivist treatment: in the perceptual case there is a hallucination (say) of a cat sitting on a mat; in the aesthetic case there is an illusion of feeling (for example) fear. Note, however, two differences between the perceptual and aesthetic situations.

First, conjunctivism conceives the structures of the two situations in somewhat different ways. In the perceptual case, the conjunctivist's highest common factor starts life, as it were, on the bad disjunct, where it is a hallucination. The idea is then that, by applying the argument from illusion, we can find the same factor on the good disjunct, but supplemented, in this case, externally by a causal story that is absent on the bad disjunct. (The absurdity of the conjunctivist's position may be expressed by noting that, on this approach, a successful perception is at least a hallucination, and then something else gets added which turns it into a genuine perception; it is at least a failure, and then something else gets tacked on to it which turns it into a success.)³⁰ In the aesthetic case things go the other way round. Here the highest common factor of the conjunctivist's story starts life, as it were, on the ‘good’—or rather the ‘real’—disjunct, where I feel (say) real fear. The conjunctivist then claims to find the same factor present on the ‘bad’—or rather the ‘unreal’—disjunct, where I have an initial feeling of (say) real fear, which is then ‘corrected’ by a superego reminding me that the situation to which I am responding is not, in the relevant sense, real, but an artistic imitation of reality. For the perceptual conjunctivist, the ‘unreal’ disjunct is simple, the ‘real’ disjunct articulated into logically discrete stages; for the aesthetic conjunctivist, it is the other way round.

Secondly, we have agreed that in the perceptual case a hallucination may, ‘from the inside’, be indistinguishable, or simply not distinguished, from a genuine perception. That cannot be so in the fictional case, because then quasi-fear, for instance, would be real fear, and the audience would react accordingly. As we have said, quasi-emotions are shot through with aesthetic pleasure; but the corresponding real emotions are, in general, not. The point about Nero's bodyguard in Suetonius’ account is that he really is afraid on his master's behalf: he has momentarily forgotten the theatricality of the whole set-up. Like Aristotle's akratic agent, he has temporarily lost the belief (in favourable cases, knowledge) that he had. Feeling a quasi-emotion is governed by the subject's belief (knowledge) that the situation in which he or she is placed is, in the relevant sense, fictional; if that belief is lost the way is open to feeling ‘real’ emotions. Equally: ‘The delight of tragedy proceeds from our consciousness of fiction; if we thought the murders and treasons real, they would please no more’.³¹ (Of course, an imitation may be indistinguishable from the real thing, but recall from earlier in this section that quasi-emotions are not imitations of their genuine congeners.)

Despite these differences, it remains true that the aesthetic scenario is similarly structured to the perceptual one in the sense that the ‘unreal’ disjunct is wholly described in terms set by the ‘real’ disjunct, which has conceptual priority. In the aesthetic case, of course, it is not appropriate to think in terms of failure versus success: a hallucination is a failure to enjoy a genuine perception, but an aesthetically pleasing stage performance is an artistic success, not a failure to be something else. Nevertheless, the parasitism which we observed in the perceptual case carries over to the aesthetic one: aesthetic representation, being an imitation of real life—that is, of life lived outside the institution of fiction—is ontologically asymmetrically dependent on that real life. (This asymmetry obtains even if, once the institution of fiction exists in a culture, members of that culture find their perceptions of reality interpenetrated by their experiences of fiction.)³² There is also, as we have observed, an analogy to the way the particular–general distinction applies in the perceptual case: hallucinatory content abstracts—generalizes—on the de re, object-involving content of the corresponding genuine perception; the sentences of fictional discourse have an underlying logical form that is general in nature; its proper names are descriptive. In general, fiction concerns itself with abstract relations and possibilities: that recalls Aristotle's point (Poetics, ch. 9) that while history particularizes, tragedy generalizes; history is about what did happen, tragedy the sort of thing that might happen. In both perceptual and aesthetic cases, movement from the ‘real’ to the ‘unreal’ disjunct involves a ‘thinning out’ of content—a replacing of particular, de re content with general and descriptive.

I have been allowing myself to characterize the aesthetic scenarios that have been my object of concern as ‘fictional’. But if we are interested in solving the paradox of tragedy we need to factor in the point that non-fictional literature, such as history, can be, and often is, tragic in the relevant (aesthetic) sense: our response to, say, Tacitus’ account of the death of Germanicus does not differ essentially from our response to Sophocles’ treatment of the fate of Oedipus. Indeed, in the hands of the great historians, ‘history might be said to be still more tragic than tragedy itself’.³³ Hume, praising Cicero's ability to rouse the passions of his audience, writes:

The pathetic description of the butchery, made by Verres of the Sicilian captains, is a masterpiece of this kind: But I believe none will affirm, that the being present at a melancholy scene of that nature would afford any entertainment. Neither is the sorrow here softened by fiction: For the audience were convinced of the reality of every circumstance. (1987: 219)

Hence, if fictionality is defined, as Hume is defining it here,³⁴ in such a way as to exclude the portrayal of real events, we should be comparing the perceptual case not narrowly with fiction but with artistic representation in general: for in the historical case, as in the dramatic, what moves us is a certain representation of the events.

Although Hume is right that fictionality in the sense of that word which excludes fact is not essential to eliciting the distinctively tragic response in an audience, it is significant that his Verres example concerns a place (Sicily) geographically distant from the Rome of his audience, just as Germanicus’ death was temporally distant from Tacitus’ readers. That remoteness is crucial in generating the tragic response: in stage tragedy a principle of éloignement has indeed typically been observed.³⁵ Distance may be achieved in various ways, by fictionality (in the sense which excludes fact) or otherwise.³⁶

Still, as Aristotle saw, the distance must not be too great: the tragic sufferers must be like us (Poetics 1453a4–7). So one might say that the tragedian aims to find the ‘Goldilocks zone’ between too much distance and too little. The continuum that stretches from too much distance at one end to too little distance at the other comprises: (i) a part at one end where the audience feels indifference; (ii) a region in the middle, where the appropriate quasi-emotion is felt; and (iii) a part at the other end, where the corresponding real emotion is felt, because here the tragedy is felt to be too close. (These ‘parts’, and the catastrophe point at which quasi-fear, say, flips and becomes real fear, are all, of course, vaguely delimited.) We might then say that what is essential to the tragic response is that either we are presented, as most stage drama does present us, with a mere imitation—a façade lacking the sort of causal background that we would expect in the non-fictional case—or if, as in the case of ‘tragic’ history, we are presented with a description (purportedly) of actual happenings, with the appropriate causal background in place, there then exists sufficient distance between the audience and the relevant events for the depicted events to be consumed as fiction. And what that means (Section VII) is that it is the generality of the subject matter that is functional in producing the felt quasi-emotions. The consumer's emotional response is generated by the descriptive form of the narrative.

In the epistemology of perception disjunctivism ensures that the mind is connected up in the right way to the experienced world. As far as concerns our reflections on the ‘tragic’ response, we need disjunctivism to ensure that mind is connected up in the right way to action, and in particular we need it to model the very different functionalities that emotion and quasi-emotion (at least in general) have. Conjunctivism, with its highest-common-factor account, splits the self. In the perceptual case it divides the mind, which in a successful perception contains the world, into what it thinks of as strictly internal and strictly external components, merely adventitiously related; the cost of that manoeuvre is a disastrous divorce between mind and world that leaves the ‘internal’ residue of the division bereft of content. In the aesthetic case conjunctivism again divides the mind, again to disastrous effect. As Wittgenstein taught us, the self is essentially an agent, and contains its own behaviour, in the sense that being in a particular mental state is logically keyed not indeed to the particular acts that are in fact performed by the agent—taken individually, those acts are contingent products of the relevant states—but to particular sorts of behaviour, or rather to tendencies to engage in particular sorts of behaviour: for these tendencies are necessitated by the relevant mental state. Feeling fear, for example, is a matter of behaving, or at least of being disposed to behave, in a certain way; it is not an ‘internal’ mental state merely accidentally related to ‘external’ behaviour. So the objection to conjunctivism is, au fond, the same in both cases: it splits the self, in the one case the experiencing self, which logically contains the world, and in the other the acting self, which logically contains its behaviour. These errors evince a common pattern, from which disjunctivism rescues us.³⁷

Footnotes

References