The Problem of Tragedy and the Protective Frame

The Paradox of Tragedy

Hume opens his 1757 essay, “Of Tragedy”: “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 and uneasy” (Hume, 1987, p. 216). How is it that such emotions as sorrow, terror, and pity—otherwise nearly always unpleasant—are pleasurable in a well-crafted tragedy? How can we enjoy something that is—seemingly in its very essence— negative?

Hume’s answer is that our pleasure arises from the artistry of the tragedy: we naturally find imitation agreeable, and in attending to a work of artistic imitation, we are disposed to agreeableness. Within this state of disposition,

[t]he impulse or vehemence, arising from sorrow, compassion, indignation, receives a new direction from the sentiments of beauty. The latter, being the predominant emotion, seize the whole mind, and convert the former into themselves, at least tincture them so strongly as totally to alter their nature. (Hume, 1987, p. 220)

As we enjoy the artistry of imitation, what we experience as a part of that imitation only adds to the enjoyment. Where fear or sorrow arising from events witnessed in the real world would be negative, when roused in a staged tragedy—being subordinate to the pleasure of imitation—the negative emotion is converted into a positive emotion, only adding to the pleasure of the experience. Fear retains its character as fear, and sorrow as sorrow; only the valence has inverted.

Hume’s theory turns on the empirical claim that audiences will experience a sort of positive fear or positive sorrow in response to a good tragic fiction. This suggests that the overall experience of those who report enjoying such fictions should be positive. Andrade and Cohen (2007) report, however, that rather than experiencing negative arousal as pleasurable, those who purposefully engage with fictional terror experience similar levels and patterns of negative feelings as those who purposefully avoid such things. Notably, in response to short clips of horror movies, participants of both groups in Andrade and Cohen’s study reported both positive and negative emotional responses (rather than the latter being converted to the former). There are, however, at least two problems with Andrade and Cohen’s experiments. First, participants only watched brief clips from The Exorcist and Salem’s Lot, but Hume’s theory contends that audiences must first be primed by the artistry of an imitative tragedy in order to convert their negative emotions to positive ones. Most fictional tragedies and horror movies do not open with scenes of horror, but take a fair amount of time to ramp up. The scene used from The Exorcist, for example, occurs nearly 100 minutes into the film. Without proper priming, Hume might suggest, it should not be surprising that conversion does not occur.

The second problem is more fundamental. Andrade and Cohen’s (2007) first experiment uses the familiar PANAS scale, which works on the assumption that fear, for example, is an inherently negative emotion (Watson, Clark, & Tellegen, 1988)—precisely the assumption that Hume rejects. If, as Hume suggests, one might experience fear, sorrow, or anxiety, but as a positive emotion, the PANAS scale would not be able to account for this. In Andrade and Cohen’s second experiment, participants reported their emotional state at any given moment on a grid with degrees of “happy/joyful/glad” running along the Y-axis and degrees of “afraid/scared/alarmed” along the X-axis. Although participants may select a spot on the graph that represents feeling a combination of happiness and sadness, the grid builds in the assumption that “happy” and “afraid” are inherently opposed affective states, an assumption that Hume (1987) rejects.

How is it, we might ask, that an essentially or regularly negative emotion like fear, sorrow, or anxiety, can be converted or tempered into a positive emotion, while retaining its recognizable character of fear, sorrow, or anxiety? The key element of these compensatory theories is not whether they preserve the fear, it is how they preserve the character of the fear.

Feagin (1983) offers a new compensation theory, contending that our displeasure is a response to the events unfolding in the tragedy, while our pleasure arises as a sort of moral response to that response: in essence, we approve of our moral disapproval of the fictional events. Hofer and Wirth (2012) appear to support Feagin’s theory. Participants watched a sad film, and were asked to assess their emotional responses to the film, the appropriateness of those emotional responses, and their overall enjoyment of the film-watching experience. Hofer and Wirth found that, where participants assessed their sad emotional responses in watching the film as “appropriate” or “adequate”—in general, as “norm compatible”—the greater their overall experience of enjoyment: “the higher the level of sadness, the greater the evaluation of the emotion as norm compatible, and in turn, the greater the experience of enjoyment” (Hofer & Wirth, 2012, p. 51). However, Hofer and Wirth fail to assess whether participants are enjoying the norm-compatibility of their responses, or whether they are enjoying the sadness itself. The first possibility is in line with Feagin’s theory, the second is in line with Hume’s, and both are in line with the data.

Another compensation account offered by Carroll (1990) shifts discussion to the contemporary genre of horror. Carroll suggests we are both attracted to, and repelled by, fictional horrors: monsters, zombies, Blaculas, and Frankenstein’s monster repel us in violating our basic categories of life and death, but intrigue our curiosity at the same time and for the same reason. Carroll (1990) writes, “The fascination of the horrific being comes in tandem with disturbance. And, in fact, I would submit that for those who are attracted to the genre, the fascination at least compensates for the disturbance” (p. 189).

Problematically, compensation accounts generally postulate either a level of conscious introspection that is not an obvious part of the ordinary tragedy or horror audience’s experience, or an unconscious phenomenon with little evidence to suggest it. Moreover, such accounts tend to leave rather gaping holes: Feagin’s account does not easily apply to fictions that centrally turn on shock (Alien) or creepiness (Uzumaki) rather than some moral transgression, while Carroll’s theory has trouble dealing with fictions like Slumber Party Massacre, that, for all their apparent horror-ness, lack the sort of category-violating being at the core of his theory. Eaton (1982) and Morreall (1985) have offered theories suggesting that our enjoyment of tragedies and horror stories relies on our ability to appropriately distance ourselves from the fictional events unfolding. Eaton writes:

A horror story is fun to read only when we are in control of the situation in which we read it. . . . Often recognition that what we are reading, watching, hearing, etc. is “unreal” is enough to assure us that we are “in control.” (p. 59)

Morreall (1985) suggests that the enjoyment of fear may reside in the body’s physiological, autonomic responses to danger. It is the excitement felt in such changes, Morreall suggests, that opens the door for enjoying fear, and it is control that keeps the excitement from becoming overwhelming terror.

The control required for enjoyment is, on Morreall’s view, a function of our desires, and rests on our capacity to start, stop, and direct our attention. This will be easiest when we are attending to something—a novel or a movie— that has no real-world practical consequences for us. So long as control is maintained, Morreall suggests, “we are able to enjoy these negative emotions, as we are frequently not able to enjoy negative emotions toward real life situations” (Morreall, 1985, p. 101). Eaton and Morreall compare the enjoyment of tragedy and horror with the enjoyment of extreme sports and roller coasters. Eaton (1982) writes: “Avid roller coaster fans are never completely unafraid, indeed . . . the feelings of exhilaration sought come only if some element of risk is involved. But a risk that we believe we can handle” (p. 59). For it to be enjoyable, the fear engendered by roller coasters and horror movies must be tempered by an at least tacit belief that one is reasonably safe from any danger—when this belief is erased, the fear is no longer enjoyable.

Gaut (2002) calls control theory “ingenious, but inadequate,” suggesting that it “leaves it utterly mysterious how the mere fact that I can choose to attend or not to an otherwise unpleasant emotion, such as fear, could render that emotion pleasant” (p. 300). Gaut suggests, instead, that:

[W]hat makes negative emotions negative is not the painfulness of either the emotional response or of the object. Rather, it consists in the fact that objects to which these emotions are directed are brought under negative evaluative concepts: the dangerous, the wrongful, the shameful, etc. (pp. 302–303)

Whether people experience a negative emotion as unpleasant, Gaut (2002) argues, depends on whether they value or disvalue the objects of those emotions. If one positively evaluates some state of affairs, it would be normal to feel pleasure at achieving it. Typically, Gaut argues, we disvalue experiences of fear and sadness, but there is nevertheless room for atypical enjoyment of such experiences. Normally, we would not enjoy being in mortal danger, or witnessing some tragedy unfold. In a tragic play or horror movie, however, the knowledge that what we are seeing is fictional allows the possibility of valuing—and so enjoying—such an engaged experience. It is not, then, that fear and sorrow are necessarily or intrinsically unpleasant, but only that they are typically unpleasant. Where fear and sorrow typically produce the desire to escape or fix the object of the emotion, we can and do occasionally engage with such emotions. There is nothing paradoxical, Gaut suggests, about desiring experiences of fear and sorrow.

That what allows us to enjoy fictional tragedies and horrors is that there never really was a problem—that fear and sorrow are not necessarily negative—presents an elegant basis upon which to dissolve the paradox. However, Gaut’s (2002) contention that pleasant experiences of fear and sadness are atypical is at odds with the sheer popularity of fictional tragedy and horror. Stephen King, Dean Koontz, and children’s horror author R. L. Stine have sold over a billion books combined. In a recent poll of respondents’ favorite plays by the Bard, The Royal Shakespeare Company found that his gruesome tragedies topped the results. Hamlet took 26%, King Lear 18%, and Macbeth was tied with The Tempest at 10% (Royal Shakespeare Company, n.d.). Our appreciation of the fear and sorrow engendered by fictional tragedies and horrors is hardly atypical; it is epidemic.

Word limits prevent us from outlining other philosophical attempts to account for our enjoyment of negative emotions in art. From the Ancients up through the contemporary era, however, theories have been moving steadily in one direction—a direction, as it happens, that is in line with theories in contemporary psychology.

Reversal Theory

Hebb (1955) outlined the foundations of optimal arousal theory, a view of motivation suggesting that for each person, there will be an optimal level of arousal for effective behavior: “up to a certain point, threat and puzzle have positive motivating value, beyond that point negative value” (p. 250). Although the optimal level of arousal varies by person on the theory, for most of us the optimum will be somewhere between low and high arousal. Increasing arousal means excitement; decreasing arousal means relaxation. However, where arousal is too low, we are bored, and where it is too high, we are anxious—at either extreme, the result is unpleasant. So, optimal arousal theory suggests an ideal midpoint between extremes, where arousal is high enough to avoid boredom, but not so high as to engender anxiety.

Although optimal arousal theory quickly gained acceptance, Apter and Smith (1977; Smith & Apter, 1975) began to question its assumptions. Apter (1976) introduced findings suggesting that the middle ground of arousal was not always the optimal state. Rather, subjects more often reported maximum pleasure when a situation was either on the very low level of arousal (having a warm bath) or on the very high level (reading a tense thriller or building to orgasm). To account for such findings, Apter and Smith developed reversal theory as an alternative to optimal arousal theory. Rather than treating emotional arousal as a continuum with pleasant affect in the middle and unpleasant at the extremes, reversal theory suggests that low arousal may be pleasant (relaxation) or unpleasant (boredom), and that high arousal may likewise be pleasant (excitement) or unpleasant (anxiety). High arousal will be pleasant when one is in an excitement-seeking mode, and unpleasant when one is in an anxiety-avoiding mode. When one is seeking excitement but fails to find it, one will be bored; when one is seeking to avoid anxiety and succeeds, one will be relaxed. When we are focused on our goals, and in what is known as a telic state, we find arousal unpleasant because it stands between us and our goals: arousal becomes anxiety. Pleasure in the telic state corresponds with perceived progress towards one’s goals. Conversely, when we are concerned with enjoying ourselves in the moment, rather than a goal, we are in a paratelic state—here, we find arousal pleasant (as excitement) and seek it out (Apter, 1992, 2007a, 2007b; Kerr, Murgatroyd, & Apter, 1993).

We encourage a paratelic state by engaging a protective frame, which deemphasizes our real-world concerns. Apter suggests three sorts of protective frame: the confidence frame (in which confidence in one’s abilities reduces the perceived threat of real danger); the safety-zone frame (in which one’s perceived protection psychologically distances oneself from some real danger); and detachment frames (in which one thinks oneself a mere observer of some danger, real or imagined). The seasoned skydiver is in a confidence frame; the amateur boxer is in a safety-zone frame; and the horror-movie-goer is in a detachment frame. Apter suggests that “within any type of protective [frame], i.e. when the excitement-seeking state is in operation for whatever reason, any arousing emotion will be pleasant—even supposedly negative emotions” (Apter, 2007b, p. 76).

Reversal theory thus offers the mechanism that Gaut (2002) found lacking in control theory (Eaton, 1982; Morreall, 1985), and further explains the mechanism of conversion in Hume’s (1987) account. According to Apter (2007a),

In watching a play or reading a novel we may experience a variety of “negative” emotions—such as anger, horror, grief, contempt, disgust—and, within the detachment frame, we enjoy them. Why else would we go to watch tragedy, horror films, family dramas, “weepy” movies, and the rest? The negative emotions in these cases, and indeed in all of fiction, are not there simply to be tolerated: they are the principle vehicles of our pleasure. Thus the best moment in the horror film is a moment of shock; in the family drama it is the moment of disclosure of a scandal which has been kept secret; in the romantic film it is the moment of pathos when the heroine dies. Without these moments when we are moved or thrilled, however supposedly distressing the particular emotion, we would feel cheated. (p. 52)

In reversal theory, these otherwise negative emotions are called parapathic emotions, “which exist alongside their natural counterparts in the telic mode, and go by the same name as their telic ‘twins,’ but which, in the paratelic mode, have an inverted relationship to hedonic tone” (Apter, 2007a, p. 53). The philosophical error at the heart of the paradox of tragedy was assuming that fear and sorrow are essentially negative emotions. Even Apter tends to treat fear and sorrow as default telic (and so, in these cases, negative) affective states. However, fear and sorrow are in themselves neither positively nor negatively valenced; rather the hedonic tone of each depends on one’s metamotivational state.

Within the detachment frame, horror-movie audiences engage in a process of self-substitution, in which one identifies and empathizes with a fictional character. In such a case, Apter (2007b) suggests, “the paradox is that one must identify and empathize closely enough with one’s ‘substitute self’ . . . to be able to experience something of the latter’s emotions yet remain detached enough to retain the protection of the frame” (p. 67). This runs into some problems, however. Carroll contends that the idea that we identify or empathize with fictional characters is conceptually confused. As an example, at one point in the movie Halloween, Michael Myers has just killed one character, Bob, and now appears disguised in the bedroom doorway, where Bob’s girlfriend Lynda waits for him. We feel that paratelic fear, but who are we empathizing with? Bob is dead, so he feels nothing. Myers is an unfeeling killing machine, who, if he did feel, certainly would not be feeling fear. And Lynda has (so far as she knows) no reason to be afraid. Carroll suggests that in most cases, we do not feel empathy, but sympathy: we feel for those characters, but not with them (Carroll, 2013). And, though Carroll has also suggested that “in order to respond appropriately to something like a horror film . . . we must believe we are confronted with a fictional spectacle,” (Carroll, 1990, pp. 67–68) sympathy seems only to require a detachment frame, not one in any way specific to fiction. Presumably, we can sympathize at least as easily with a real person as with a fictional character. Apter has suggested that one can “watch the evening news on television as if it were fiction” (Apter, 1993, p. 32). Presumably, however, to watch the news as if it were fiction would be to treat it as fiction: as something one disbelieves—and certainly this is not normally the case. Rather, one can (and often does) watch the news in the same sort of detached mode in which one engages with fiction. Put simply, when it comes to emotional engagement, the important division is not one between fiction and nonfiction, but between being attached and being detached (and one can be detached from either a fiction or a nonfiction). McCauley (1998) suggests that what is central to the enjoyment of horror movies is the “dramatic distance—framing violence as fiction” (p. 144). McCauley suggests that the music and voiceovers in horror films signal the unreality of the films, allowing viewers to frame what they are seeing as fiction, and so allowing for enjoyment. However, all of these elements are likewise present in documentaries (e.g., Taxi to the Dark Side), docudramas (e.g., 127 Hours), and horror movies actually or purportedly “based on a true story” (e.g., The Amityville Horror), each with the same power to evoke enjoyable fear and sadness as their fictional counterparts despite being treated by viewers as the recounting of real events. Certainly, disbelief can encourage or bolster a protective frame, but such disbelief does not seem to constitute an element of the frame itself. The frame is the same. Maintaining the detachment frame does not centrally depend on whether one takes a story to be fictional, but whether one takes the events of the work—or the work itself—to affect one’s real-world concerns.

Some empirical work has been done to test the power of protective frames and psychological distance. Working with five studies, and factoring over several forms of psychological distance (spatial, temporal, hypothetical, etc.), McGraw, Warren, Williams, and Leonard (2012) conclude that “psychological distance increases the humor perceived in the averse, severe violations (i.e., tragedies), but that closeness increases the humor perceived in less aversive, mild violations (i.e., mishaps)” (McGraw et al., 2012, p. 1221). Although tragedy is mentioned here, what the authors label as a tragedy is the case of a woman unknowingly donating $2,000.00 via text messaging. As this is not the sort of “tragedy” at the center of the paradox of tragedy, however, and as the concern in the paradox is not with humor but rather enjoyment of typically negative emotions, the study is only suggestive for our purposes. In a third experiment in Andrade and Cohen’s (2007) study, the authors repeatedly reminded some participants that the characters in a horror film were simply actors playing roles, and found that such a protective frame allowed for “coactivation” of “feelings of opposite valence (e.g., fear and happiness)” (p. 294). Even Andrade and Cohen, however, assumed that fear and happiness are of opposite valence.

One of the central goals of analytic philosophy is to question those assumptions in other arenas of study—and in philosophy itself—that tend to go unquestioned. Most extant empirical studies on the topic at hand have focused on “mixed emotions”—experiencing both positive and negative emotions in response to some stimulus: the “concurrent experiences of a positive and a negative emotion that preserve the quality of the two emotions as opposed to creating an entirely new quality” (Hemenover & Schimmack, 2007, p. 1103). However, as discussed before, these studies tend to assume (and employ metrics that build in the assumption) that such emotions as horror, disgust, and sadness are inherently negatively valenced—not things one might enjoy. However, if such emotions as fear and sadness are not essentially negative, then empirical tests that build in the assumption that they are would be grounded on a faulty assumption, with the error being built into the very questions. Now, how much this affects the veracity of such tests would be nearly impossible to predict. In fact, it may be the case that this assumption merely shapes the language used to discuss emotions. Further complicating the problem is that this notion of fear, etc., being essentially negative is also likely built into our more widespread folk psychology. That is, most nonpsychologists likely take this for granted as well, and so any metric which seeks to eliminate this assumption is going to take some care to build. Since Apter’s focus is not on mixed emotions but on the way in which a frame can make a negative into a positive, reversal theory fits nicely with the assumptions and language of the philosophy literature where the negative emotion itself become pleasant.

While Apter is generally careful to avoid treating fear as negative by default, he is less careful to avoid treating telic metamotivational states as the default. Talk about protective frames as vehicles people use to leave telic states and enter paratelic states suggests that equilibrium is found in the telic state, and that we must take action to enter the paratelic. The implication, then, is that the telic state is itself frameless, a matter that is not altogether clear. Further investigation is needed on those things that encourage detachment (the protective frames) and those things that encourage attachment (as yet unnamed, but we like “attachment frames”).

Closing Remarks

Obviously, we have been able to do little more here than give a sketch of how Apter’s theory, which is steadily gaining ground in psychological circles, offers great promise for resolving the classic paradox of tragedy as an acute form of the more general paradox of mixed emotions. Indeed, Apter’s theory goes a long way towards explaining the intuitions of a number of the philosophers we have discussed here—and many others besides—while avoiding their central problems. Philosophical insight also has much to offer in terms of refining the details of reversal theory and the mechanics of the protective frame. Despite diverging in the Modern era, it seems that philosophical and psychological theories have again begun to converge. However, while recent work in philosophy and psychology appear to be running on parallel tracks, they have so far remained at arm’s length. It seems that a bridging of the gap between the domains offers the greatest promise for solving this and countless other perennial puzzles. The problem of tragedy and the protective frame seem an ideal starting place to build a bridge.