2. Literary fiction

In the context of literary fiction, the report model thesis is equivalent to the thesis that, by default, stories have internal narrators—narrators who belong to the world of the story. For if the report model thesis is true, and so one is to imagine learning various facts from someone’s report, then that someone is an internal narrator; and, conversely, if a story has an internal narrator, then one is to imagine that the details of the story have been reported to you as fact by that narrator.Footnote ² Arguments for the ubiquity of internal narrators are, therefore, arguments for the report model thesis. A first such argument is due to Kendall Walton:

Granting the last point for the sake of argument—that unavoidable imaginings are required—, we may dispute the first, that these imaginings are unavoidable or inevitable. In many contexts declarative sentences are not used to report events or describe situations, and we know this. When the handwriting teacher says, “using good cursive, write in your notebook that snow is white,” there is no temptation to think that the embedded declarative, “snow is white,” is being used, by the teacher or student, to describe a situation. Similarly, for the known falsehood “the moon is made of green cheese,” in the instruction “present at least five arguments against the thesis that the moon is made of green cheese.” The context is one in which something else is to be done with the sentence or its content. It may be that the declarative sentences making up a story occur in a similar context, and we know this. For example, those sentences may, by convention, all be embedded under an unwritten imperative: “Image the following…” If they are, and we know they are, we will not project the standard use of declarative sentences onto this case; we will not imagine that someone has used the sentences we read to describe anything.Footnote ³

For a second argument, Derek Matravers defends the report model thesis for literature by attacking the transparency model in a standard case:

True, visualizing Emma brushing her hair would make it hard, if not impossible, to simultaneously read some words on a page. But the transparency model does not require this; Matravers misunderstands the model. It tells us to imagine that Emma groomed herself, but it does not tell us to imagine our coming to know this by any particular means, certainly not by seeing it. And imagining (simply) that Emma groomed herself does not interfere with continuing (actually) to read the words on the page.

The third, and strongest, argument for the report model to be considered here points to a phenomenon that hypothesizing an internal narrator could explain. “In” a story, one often detects attitudes toward its characters or events. Pride and Prejudice has a great deal of sympathy for Elizabeth Bennet, while Charlie and the Chocolate Factory holds all the children save Charlie in contempt. But in its primary use, “holding in contempt” denotes a mental act (similarly for “has sympathy for”). Since stories lack minds, the phrase can only apply to a story in a derivative sense. In need of answering, then, is the question of what it is in virtue of which Charlie holds the children in contempt. The best answer, the argument goes, is that the story holds the children in contempt in virtue of its (otherwise invisible) narrator holding them in contempt (Walton, Reference Walton1990, p. 366).

But the hypothesis that a story has an “effaced” narrator whose presence is never felt, and whose only job is to be the mind in which certain attitudes reside, is on its face implausible. Moreover, it is unclear whether the hypothesis really explains the phenomenon. If a story has a narrator, why should the attitudes of the narrator automatically become the attitudes of the story? After all, the narrator is a character in the story, and no other character’s attitudes are automatically those of the story; there is a gap between Charlie and the Chocolate Factory’s attitude toward Augustus Gloop, and Augustus’s own. If a similar gap between the narrator’s attitude and the story’s is impossible, we are owed a reason why. Finally, when a story has an overt narrator, there is often a gap between the story’s attitudes and the narrator’s: Huck Finn and Holden Caulfield are canonical examples. In these cases, some other account of the story’s attitudes is needed; simplicity considerations favor extending that account to stories without overt narrators.

3. Cinema

In the context of cinema, a core argument against the transparency model thesis is that it cannot tell the difference between reading a novel and watching a film. The Godfather exists in both novel and film form. If appreciating the novel and appreciating the film alike involved simply imagining that Don Corleone did this and that, that Michael Corleone avenged the attack on his father, and so on, then the two acts of appreciation would be the same. Since they are not, another model is required.

In this case, the argument fails because the novel and the film differ in many details. In the transition from novel to screenplay to shoot, cuts and revisions were made. What, though, about a (hypothetical) perfectly faithful film adaptation of The Godfather, one whose fictional world was identical to the fictional world created by the novel? Now surely the imaginings mandated in the two cases by the transparency model are the same. But surely there is a difference.

The report model is supposed to succeed where the transparency model fails. On that model, one is not merely to imagine that Michael Corleone avenges the attack on his father, when watching the film. One is also to imagine learning of his vengeance in a particular way: by watching a cinematic report of it. When one reads the novel, by contrast, one engages in no such imagining.Footnote ⁴ This argument, if good about this case, supports the report model thesis for cinema.

The report model is not the only alternative to the transparency model in the context of cinema. Also available is the perceptual model. That model tells one to imagine seeing Michael’s vengeance itself, not a cinematic report of it.Footnote ⁵ The perceptual model thesis says that imagining seeing the characters and events is the default. The focus here will be on the report model thesis; analogs of the arguments below can be made against the perceptual model thesis.Footnote ⁶

If, as the report model thesis says, viewers of a film are to imagine seeing a cinematic report of the story’s characters and events, then what are they to imagine about how the report came to exist? One answer is that the characters were followed around by a documentary film crew, one with the power to make themselves invisible to their subjects (who never notice them), and the power to position cameras in seemingly impossible places (say, the mouth of an erupting volcano, hot enough to melt any mental or plastic known to man). But it cannot be that we are supposed to imagine this, since few if any viewers actually do imagine it (Carroll, Reference Carroll, Carroll and Choi2006; Currie, Reference Currie1991, Reference Currie1995). Nor do viewers imagine any alternative answer, including Wilson’s, that the imagined cinematic report that we imagine viewing is a “naturally iconic image,” not produced by any intelligence (Wilson, Reference Wilson2011).

The argument against the report model thesis here is, first, that if we are to imagine we are seeing a cinematic report, then we are to imagine the report being produced in some way or another; but second, for any such possible way, we are not to imagine that the report was produced in that way. One might reject the first premise: viewers are to imagine seeing a report but not to imagine anything about how the (imagined) report came to exist (Walton, Reference Walton2008). After all, every imagined scenario is, to some extent, indefinite. Readers of Sherlock Holmes novels are to imagine that he has hair on his head, but not to imagine that he has some particular number of hairs.

While we are not supposed to imagine that Holmes has one million hairs, or one million and one, and so on, we are supposed to imagine whatever is common to every possible answer to the question of how many hairs he has. So if, on every answer consistent with the rest of the story, Holmes has a decent amount of hair, then we are to imagine that he has a decent amount of hair. Similarly, even if, on the report model, we are not to imagine the report being produced in some particular way, it remains true that we are to imagine whatever is common to the possible ways it could have been produced. The problem is that each of them involves a strange and seemingly miraculous process. If it was not made by a documentary film crew with almost supernatural powers, then it is a naturally iconic image that was caused, in ways unknown to our world, by the events it depicts (the alternatives to these two hypotheses are no better). But just as viewers do not imagine that they are seeing a cinematic report produced by a documentary crew, they also do not imagine that they are seeing a cinematic report produced by some seemingly miraculous but otherwise unspecified process. And if they do not imagine that, it cannot be that they are called on to imagine it. The report model is false.

A last line of defense could reject the claim that viewers are to imagine whatever is common to every possible answer to “how was the report produced?” Instead, they are to imagine this, if the question of how the report was produced arises or occurs to them (as part of their imaginative engagement with the film). But this question is not supposed to occur to viewers, or is supposed to be ignored by them; the provenance of the report is not a focus of the story. On this view, the question of where the report came from is what Kendall Walton calls a “silly question” (Walton, Reference Walton1990, p. 174ff).

Now, it may be that no theory of fiction and fictionality can avoid calling some questions silly. Even still, theories that reduce the number of silly questions are, in that respect, superior.Footnote ⁷ And the silly questions thrown up by the report model do not arise at all in the transparency model.

Even if a way around this argument can be found, it remains true that the report model is under-motivated. The argument above, against the transparency model, has a false premise: that the transparency model mandates just the same imaginings for the novel and the film versions of some story, like The Godfather. The film version depicts, and so makes fictional, very fine details about the contours of Don Corleone’s face, the comb of his hair, and the cut of his suit. Watching the film involves imagining those details. Since none of them are in the novel, one does not imagine them when reading. Even on the transparency model, the imaginings in the two cases are different.

It will be replied again that this difference disappears when we consider the novel and its perfect film version. But in fact, a perfect film version of a novel—one that makes exactly the same things fictional—is impossible. This is down to the difference in representational capacities of language and image. The visual appearance of a character, as presented in words, is too impoverished to be put into an image. Images are rich in content, and there is a level of richness they cannot fall below; this is in the nature of the medium (Hopkins, Reference Hopkins1998). The descriptions of, for example, a character in a novel, fall below that level.Footnote ⁸ A depiction of a dog, no matter how thin its content, has truth conditions far exceeding that of the sentence “there is a dog.” As Currie puts the point,

The argument does not need the premise that no description could have the content of an image. Even if this were possible, the description would be very long, so imagining that content in response to reading the description would remain very unlike imagining it in response to viewing the image. The image’s content can be taken in, and then imagined, all at once, while taking in the corresponding description would take years, if not an eternity, and would have to be imagined piecemeal.

4. Comics

In literary fiction and in cinema, the transparency model thesis holds up. As a reminder, the thesis is not that appreciating a story never involves imagining anything about oneself, or imagining that one has acquired information from some source in some way. It only says that, by default, one does not do these things; when special cues or techniques are in use, one may.

Is there a medium where the transparency model thesis fails, and the report model thesis is superior? I think comics is such a medium. Again, all should agree that special cues or techniques can trigger the report model in particular cases: Deadpool, for example, addressing the reader directly. The claim to be defended is that even in comics without such cues, the report model is correct; it is the default. Important for the argument are three common features of comics: the division into panels, the use of both text and image, and the way time may work within a single panel. These features present challenges to a reader that are not present in either literary fiction or in cinema. They are challenges of integration.

In Understanding Comics, Scott McCloud waxes poetic about the importance and the power of the gap between panels—the gutter. “Comics panels,” he writes, “fracture both time and space, offering a jagged, staccato rhythm of unconnected moments.” This gap must be bridged in an imaginative act by the reader; the reader must “connect these moments and mentally construct a continuous, unified reality” (McCloud, Reference McCloud1994, p. 67). He calls this process “closure.” But even in novels and in films, what is on the page or on the screen is never the complete story. The reader/viewer must always augment it in imagination, to construct the whole fictional world. Sherlock Holmes is human, even if the novels never come right out and say it; this is an extrapolation the reader must make. Examples like this are familiar in the literature on truth in fiction (e.g., Lewis, Reference Lewis1978). Is there anything special here about comics?

McCloud says that with film and television, closure is “continuous, largely involuntary and virtually imperceptible,” but with comics, it is “anything but involuntary” (p. 68). This is a key difference that deserves elaboration. Watching a “normal” film—not something experimental or overly ambitious—one rarely needs to pause and consciously try to figure out what to imagine, from one moment to the next. This is so, even when the film cuts between points of view, or from one location to another. The same goes for reading a novel. But in a comic, this kind of conscious effort is common. It is most common in comic books with multi-panel displays, like the one in Figure 1, from The Long Halloween (Loeb & Sale, Reference Loeb and Sale1996).

Figure 1.

Although the default assumption—that the panels on this page are to be read left to right, top to bottom—is correct, one still must verify that it is true. The way the panels overlap creates some uncertainty. The panel depicting Catwoman’s leap hovers above and partly covers the background panel where she and Batman confront each other; the reader must figure out that the first of these depicts later events than the second (background), and does so from a different point of view; and the reader must also grasp that it is of no significance for the world of the story that one panel obscures part of another. As a result, the sequence of events constituting this encounter between Batman and Catwoman is not seamlessly available as it would be in a film version of the scene. Jared Gardner nicely makes a similar point, emphasizing

Related to this, Pratt emphasizes that “readers of comics are largely responsible for their own rate of processing the narrative,” in a way that viewers of cinema, with its “relentless twenty-four frames per second,” are not (Pratt, Reference Pratt, Meskin and Cook2012, p. 160; Reference Pratt2009, p. 109). A reader’s “eyes and mind play over the succession of panels at the reader’s own speed” (Pratt, Reference Pratt2009, p. 109). And that play is not invariably in the forward direction; readers “can freely direct attention back and forth among [a comics’] images and events” (Pratt, Reference Pratt, Meskin and Cook2012, p. 160).

I also mentioned the use of text. When a panel contains text, one must pause to read it, and then integrate the content of the words with the content of the images. This takes time—more time, in fact, than is represented in the panel being read. The same goes for other conventionalized, non-depictive elements in a panel, like stars circling the head of a character indicating dizziness, or lines streaming behind, indicating motion. (Mort Walker calls these two elements “emanata” and “hites”; see Walker, Reference Walker1980, pp. 28–30.)

A third feature of comics to attend to is the slippery working of time within a panel. McCloud presents a wonderful example (Figure 2, from McCloud, Reference McCloud1994, p. 95).

Figure 2.

The single image in this panel embraces a significant stretch of time, and indeed, the events depicted on the left of the image are not simultaneous with the events depicted on the right. Reading the panel, therefore, is not a simple matter of grasping what it depicts, and then imagining that content. It is like a montage; there is no single thing that it depicts as a whole. In this case as well, then, one is not effortlessly aware of what is happening in the world of the comic; time and some deliberate effort are required to sort it all out.

How do these facts about comics support the report model thesis over the transparency model thesis? My main premise is that, generally speaking, if one engages with (reads, watches, etc.) a story for some stretch of time, then, normally, one’s imaginative efforts are continuous. Support for this premise comes from the analogy between appreciating fiction and playing games of make-believe.Footnote ¹⁰ Both involve the imagination: imagining what is fictional, in one case, and imagining what is true in the game, in the other (that you are the queen, for example, or that I have just blocked your sword with my shield). It is evident from the observation that games of make-believe are played as continuously as possible. Children do pause their games to negotiate the rules or settle disputes, but there is pressure not to do this. Ideally, the game’s imaginative project should continue uninterrupted. I suggest it is the same when appreciating with fiction: other things being equal, the preferred model of appreciation is the one that secures the greatest continuity in the reader/viewer’s imaginative efforts.

Now, the transparency model, applied to comics, says that we imagine the features of the world of the comic given in the panels—in the images, text, and other features. (We also, of course, imaginatively fill in the gaps between the panels.) But if that model is correct, one’s imaginative project will be constantly interrupted. One inspects a panel and apprehends what information about the fictional world that panel conveys, that such-and-such; one imagines that such-and-such; and then, typically, one needs to pause, figure out which panel is next in the sequence (again, this is not always obvious); then apprehend how it continues the story from the previous panel. If one, in Pratt’s words, “directs one’s attention” back to a previous panel, the transparency model says we have re-wound and re-started the fictional world at an earlier time—something we cannot easily do with cinema (and then only at home), and rarely do when reading literary fiction. Finally, while looking at a single panel, if it contains words or other non-depictive elements, one needs to figure out how they relate to the events the panel depicts (are they in a word bubble? a thought bubble? or something else?). Only after all this cognitive activity is complete can you resume the imaginative project, and imagine what you have apprehended. (In extreme cases, the cognitive activity can be quite taxing and take quite a while: in Here by Richard McGuire, adjacent panels jump back and forth between decades and even centuries.)

What does the report model say about reading a comic? It says that what one imagines, in the first place, when reading a comic, is that one is reading a comic, just a non-fictional one, purporting to report on actual people and actual events. In this model, then, the pausing and the effort exerted as one’s attention shifts from one panel to the next; the glancing back at earlier panels, or ahead to later panels, before proceeding; and the integrating of text and image, are of course still there. But now, doing those things does not require pausing and re-starting one’s imaginative engagement with the comics’ fictional world. Instead, one incorporates those activities into one’s imaginative project; in doing those things, one also imagines doing them.

It might be objected that what we are asked to imagine, on the report model, is something of which we have no conception. If one does not know what a non-fiction comic book might be, one will not know how to imagine, of a fictional comic book, that it is one. Panels, cartooning, speech bubbles—do we really have a sense of how these devices could be put in the service of non-fiction? But this objection is easily answered: non-fiction comic books exist, and even top best-seller lists—Kent State by Derf Backderf, for example, or Fun Home by Alison Bechdel. Those who buy them do not find them strange or hard to understand.

5. The illusion of depth, cartooning, and a challenge

Figure 1 above illustrates another piece of evidence for the report model thesis for comics: the “illusion of depth.” The panels overlap each other, as if arranged in three-dimensional space. Gavaler notes that this is a common technique:

When this technique is in use, as in Figure 1, it is natural to imagine that the illusion is real: that one is, in fact, looking at panels arranged in three-dimensional space. But to imagine that is to imagine in accordance with the report model.

The default artistic style for comics is cartooning: simple line drawings that, when drawings of human beings, resemble actual humans only at a quite abstract level. That this is the default further supports the report model thesis for comics. The transparency model is most plausible for some given work when the facts about the work’s fictional world that one is to imagine are available relatively immediately, when one reads or looks at the work. Reading “she pulled down her spectacles” in a story, not much thought is required to know that one is to imagine that she pulled down her spectacles. Seeing images of Buster Keaton chasing a train, one knows right away that one is to imagine Johnny Gray (his character in The General) chasing a train. But reading a Calvin and Hobbes strip—of course, one knows one is to imagine Calvin doing this or that, but what is one to imagine about Calvin’s appearance? Surely, the panels of the comic are not photorealistic depictions of Calvin. Presumably, one is to imagine that Calvin is a normal-looking human boy, who’s appearance is aptly simplified or caricatured by the images on the page. But it may sometimes, or often, be unclear when some aspect of Calvin-as-drawn is just part of the caricature, and when it corresponds to an actual feature Calvin has. This is another of those moments of potential uncertainty that, if the transparency model is correct, must be a moment of potential interruption of one’s imaginative activity. On the other hand, if the report model is correct, no interruption is needed: one imagines the whole time that one is looking at caricatures of a boy. One may, of course, if the report model is correct, imagine wondering which features of the drawings are distortions and which are accurate; but this imagined wondering is, obviously, no interruption of the imaginative project that reading the comic involves.

My argument assumes the falsity of the “panel transparency thesis,” which says that characters in comics are as they are depicted in the panels. (Note that the panel transparency thesis is distinct from the transparency model thesis, though the former thesis, if true, would make the second more plausible.) Roy Cook defended the panel transparency thesis by asserting that its falsity leaves us unacceptably ignorant of the physical features of the characters in comics: “if the characters in a comic do not appear as they are depicted in typical panels, then what do they do look like” (Cook, Reference Cook2012, p. 135)? This, however, is not a hard question. We do not know exactly what they look like, but we know that their appearance is aptly cartooned as the artist drew it. More interesting is another argument he gives:Footnote ¹¹ even if, “when confronted with a Tim Sale panel where the Batman punches a six-inch-toothed Joker,” we “affirm that the Batman struck the Joker and deny that the Joker has six-inch teeth,”

I, however, take the difficulty of making these distinctions, not to support the panel transparency thesis, but to show that its denial is best combined with the report model for comics. For nothing is problematic about imagining looking at a caricature, and imagining finding it unclear which of its features are stylistic.Footnote ¹²

A challenge for the report model thesis for cinema was that, if it is true, then many films raise hard or unanswerable questions about how the cinematic report came to be.Footnote ¹³ The report model thesis for comics faces a similar challenge. A comic panel may depict a character alone, and words in the panel may express the character’s secret inner thoughts. If we are to imagine learning about these things by reading a non-fiction comic, then what are we to imagine about how that comic came to be? If it was written by someone (if it was not directly caused, in some mysterious way, by the events depicted), how did that person come to know the character’s appearance and thoughts at that moment? It may be said, again, that these are silly questions: if they lack answers, and if it would be troubling if similar questions about reality lacked answers, nevertheless when reading a comic, the questions are to be ignored or set aside. But still, its generating silly questions was earlier held to be a weakness of the report model thesis for cinema; this must, therefore, also be held a weakness of the report model thesis for comics. There is, however, a difference. The report model thesis for cinema was ultimately rejected because there was not sufficient countervailing evidence supporting it. No support, for example, could be found in the need to distinguish cinema from literary fiction. In the case of comics, on the other hand, there is support on the other side—the argument from the continuity of imaginative engagement, given above. This support is strong enough to make the generating of silly questions acceptable.