On the other hand, my colleagues are right in their skepticism about some universal notion of "history" and "fiction," because on the practical level there is really very little universality to our enduring quest for truth. The criteria and definitions of truth shift on every imaginable level, cultural, contextual, and personal—they have to shift, in fact, if we consider this process from the perspective outlined by Cosmides and Tooby. If our metarepresenting mind is constantly busy "monitoring and re-establishing the boundaries within which each representation remains useful," then our universal "quest for truth" is really a universal quest for temporary, local, intensely contextual truths that are reliable only within "the envelope of conditions to which [they are] applicable."11 This is to say that the constantly changing boundaries and definitions of truth are not the casualty of the social-historical change but rather the key condition of the functioning of the human brain. By adjusting and redefining what constitutes the "truth" at every new social, cultural, and personal junction, we exploit, build on, develop, fine-tune, struggle with, tease, and train a broad variety of cognitive mechanisms underlying our evolved metarepresentational capacity.12
This constant hunt for truths supposes a constant delicate interplay between energy costs and benefits. Our brain is a very "expensive" device: compared with muscle tissue, it consumes sixteen times as much energy per unit weight. Monitoring and reestablishing the boundaries for truths is crucial for our existence, and yet, in some specific situations, doing it again and again on the same material could become too costly.
For example, it may be that once readers have decided on the relative truth-value of a complex cultural artifact, such as Robinson Crusoe, or, to put it differently, once they have integrated it with a relatively weak metarepresentational tagging (as a "true story"), they may experience a broad gamut of negative emotions, ranging from disappointment to anger, when they realize some time later that they have to expend more cognitive energy on drastically reassessing their initial valuation and on reintegrating Robinson Crusoe with a very strong metarepresentational tagging (as a "feignd" story) instead. Some readers may be more amenable to this kind of reassessment, which involves revising numerous knowledge databases affected by the initial processing of the story, whereas others may find this call for the extra expenditure of mental energy irksome.
Of course I am speculating here, but the question that I am grappling with is a serious cognitive issue that has to be addressed and cannot be simply dismissed. On the one hand, we can say Defoe's readers were unhappy about his having exploited their strong preference for what Cosmides and Tooby call "urgent news"—the "true narratives about relevant people and situations." But, on the other hand, think about it: why, in principle, should readers feel so angry about realizing that a story about a person whom they have never met is really a story about a person whom nobody has ever met, especially since it contains so much otherwise true and useful information?
Similarly, a cost-benefit analysis may enter into a discussion of the tenacity with which our bookstores cling to the separation between "fic
5: "Fiction" and "History
tion" and, say, "history," in their shelving practices. Can it be that imperfect as it is, this separation saves the customers a significant cognitive effort of "deciding" (subconsciously, of course), when they begin to read a book, how much of metarepresentational tagging each little element of the story will need? Once a book is placed on the "fiction" shelf, the decision about its overall truth-value has been made for us, so to speak.13 We have the cognitive luxury of knowing, as we pick up such a book, that the story it contains is, as a whole, a metarepresentation that needs to be stored with a permanent source tag pointing to the author. We can then enjoy it as such, processing some constituent parts of it with a much weaker or no metarepresentational framing at all (including the parts that conform to our general knowledge and the parts that have a real emotional effect on us and/or teach us important life lessons).
Compare the experience of picking up a book from a shelf labeled "history." We open such a book with a subconscious expectation that as a whole it could be assimilated with a much weaker metarepresentational tagging than a book from a "fiction" shelf. Of course we can change our mind in the process of reading and decide, for example, that the given treatise contains more propaganda than accurate historical information and thus store it with strong metarepresentational tagging. But again, the preliminary cognitive work has been done for us (or is claimed to have been done for us) by the publisher, who has provided enough external markings to alert us to the intended truth-value of the book, and by the bookstore's clerk, who has put it on the designated shelf.
Furthermore, once we begin to think of how cultures satisfy, reinforce, struggle with, and manipulate our cognitive predispositions, such as our constant monitoring of the boundaries of truth, we may realize, for example, that there is something deeply paradoxical in the position of historian both today and in the time of Thucydides. On the one hand, a historian strives to diminish the amount of metarepresentational framing that her readers would deploy in assimilating her book, which, taken to its logical extreme, means removing herself from readers' consciousness altogether. The ultimate goal of the historian is to have her readers store the information that she provides simply as "X," and not as "Thucydides says that 'X,'" or as "Linda Colley says that 'X.'" On the other hand, the historian's own personality (e.g., her academic degrees, her other books, the names of publishing houses that she associates with) becomes an important factor in persuading the reader that the information contained in her book has a high truth-value, that is, that it should be assimilated with a relatively weak source-tagging.
Thucydides thus had to puff himself up and put down his competitors as liars and myth-peddlers (all the while claiming that his work "is no mere piece produced for a competition"14) in order to disappear from his work, that is, to encourage his readers to perceive a historical account penned by Thucydides as simply "the" historical account or "the everyreasonable-person's" historical account. The martyrdom of the three Chinese historians bound them inextricably to Zuozhuan and as such contributed to making Zuozhuan an infinitely more trustworthy, that is, low-on-source-tagging, book. The concept of "Death of the Author" sounds titillating precisely because it is really not that cognitively feasible (i.e., there is always an author behind a. fictional text, even if her name is lost to us); by contrast, the concept of "Death of the Historian" sounds rather unexciting because the expectation of the historian's fading-out (and I don't mean physical annihilation) is implicitly built into each historical account aspiring to the high truth-value.
The phenomenology of source-monitoring may sound complicated and look complicated, but don't let that fool you: it really is complicated. The scope of issues raised by introducing the concept of metarepresentationality, as defined by cognitive psychologists, into literary and historical studies can be truly staggering. Our evolved cognitive ability to store representations under advisement; to reweigh their architectural "truth"; and to refocus our attention on a source of a given representation in proportion to our intuitive perception of that representation's relative truth-value, structures an untold variety of cultural practices.
Of course, we are still quite a way off from figuring out what is actually going on in our brains/minds when we discriminate among the levels of truth-value associated with a given representation, such as Pride a?id Prejudice; Zuozhuan; or, for that matter, a toothpaste commercial—that is, when we somehow decide on the relative truth-value of the representation as a whole and on the relative truth-value of its components. For the purpose of the present discussion, however, we can agree on the following pragmatic observation. Our cognitive makeup allows us to store a given representation with a very strong, perhaps permanent, source tag (e.g., Beowulf will always remain a story "feign'd" by somebody, and so will Pride and Prejudice). Once we are decided on the overall metarepresentational framing of the given story (a decision
mediated by a variety of cultural institutions), we can process its constituents as so many architectural truths, including the truths about emotions experienced by its characters and our own feelings in response to their emotions. The next section will thus take for granted the larger metarepresentational status of the given
6: Tracking Minds in Beowulf
literary text and focus on the interplay of metarepresentations and variously weighted architectural truths within its fictional world.
|6 V
TRACKING MINDS IN BEOWULF
ecause this book will eventually focus on several particular instances
of the novelists' experimentation with our metarepresentational ability, here is an important point worth repeating and worth clarifying. Throughout my argument, I frequently say that this or that fictional text experiments in a certain way with our Theory of Mind and/or our metarepresentational ability. What I do not want you to infer based on such statements is that I believe that some texts do tiot experiment with these cognitive predispositions. They all do insofar as every single act of writing and reading fiction deploys our ToM, and the overall cognitive outcome of such deployment is never fully predictable. Thus when I refer to Woolf's or Richardson's or P. D. James's experimentation with their readers' ToM and/or metarepresentational ability, what I really claim is that they push to their limits certain aspects of the general, constant, ongoing experimentation with the human mind that constitutes the process of reading and writing fiction.
To illustrate my point about this constant experimentation, let us turn to the text that I used in Part I as an example of a work of fiction that does not (and perhaps could not, due to material realities of its time's textual reproduction) play with multiply embedded levels of intentionality the way Woolf's Mrs. Dalloway does—an Old English epic Beowidf. Beowulf may never be able to embed more than three levels of intentionality, but it still engages our Theory of Mind in ways that vary—within certain parameters—from one moment to another and from one reader to another.
When the protagonist of the poem, the great Geat hero, Beowulf, first arrives to Heorot to save it from the terrible monster Grendel, he is taunted by one of the local men, Unferth, who (we infer, using our ToM) must be jealous of the attention and respect heaped on the newcomer. Later, however, after Beowulf has defeated Grendel and starts preparing for the fight with that monster's vengeance-breathing mother, Unferth lends him a powerful weapon, "a rare and ancient sword named Hrunting" (64).
Because the anonymous author takes particular care to inform us that it was Unferth and not some nameless drunkard in the crowd who ridiculed Beowulf about his presumed past misadventures, the present act of sword-giving can be read as a sign of a grudging change of heart in the character who initially envied the Geat hero and distrusted him.
Here is why I see this as an example of the text's experimentation with our metarepresentational ability. On the one hand, the poem clearly relies on our tendency to monitor sources of representations (e.g., "It was Unferth who initially said that Beowulf was a loser"). Similarly, the general thrust of our interpretations of Unferth's change of heart is guided by the text's emphasizing some social nuances more than others (e.g., "It looks like Unferth is somebody whose opinion will not necessarily carry the day, but it won't be completely ignored, either"). On the other hand, within these constraints, the exact effect that each particular instance of source-monitoring may have on the reader's understanding of the text remains unpredictable.
For instance, there is no certainty that when, utilizing my ToM and my metarepresentational ability and gauging Unferth's relative social importance, I register his new attitude, my reaction to his behavior will be the same as yours, or even the same as my own after five minutes of thinking further about the poem.1 I can say that Unferth is a good guy who has been led astray by drink and then came to his senses; OR, that he is a calculating fellow who can see in which direction the wind is blowing after Beowulf's first victory, and wants to be in the good graces of the winning warrior; OR, that he is one of the rare characters in the poem who actually intuits Beowulf's vanity but is powerless to do anything about this intuition, for Beowulf is destined to live out the flaws of his character, now gloriously, now tragically.
In other words, our interpretation of Unferth's and Beowulf's behavior and personality will certainly be structured by our metarepresentational ability: for the poem calculatedly feeds this ability by implying, first, that there is an important difference between the states of Unferth's mind then and now and, second, that Unferth's opinion matters to a certain degree within the social world of the poem. Still, the exact effect of this exploitation of the particular cognitive capacity remains dependent on the state of mind of the specific reader in the specific moment.
Thus any fictional gestalt (or, were I to broaden this discussion, I would say any utterance2) that deploys our Theory of Mind and/or our metarepresentational capacity experiments with these cognitive adaptations insofar as the effect of such a deployment on the reader is never fully
7: Don Quixote and His Progeny
determined: it differs from one mind to another and from one mind now and five minutes, five hours, or five years from now. Any fictional text is profoundly experimental because the brain that interacts with this text is a dynamic system. (Hence, perhaps, the pleasures of rereading: no two close encounters with the same fictional text are ever truly the same, for the brain that responds to the text changes ever so slightly with every thought and impression passing through it.)
7
DON QUIXOTE AND HIS PROGENY
lthough all fictional texts rely on and thus experiment with their
readers' ability to keep track of who thought, wanted, and felt what and under what circumstances, some authors clearly invest more of their energy into exploiting this ability than others. Indeed, we can speak of several overlapping and yet distinct literary traditions built around such exaggerated engagement with our metarepresentational capacity. In the rest of this book I focus on two such traditions: one, exemplified by the story of Don Quixote, is the subject of Sections 7—11; another, exemplified by detective stories, will be dealt with in Part III (Sections 1-4).
From a cognitive psychological point of view, Cervantes's protagonist suffers from a selective failure of source-monitoring. He takes in representations that "normal" people store with a restrictive agent-specifying source tag such as "as told by the author of a romance" as lacking any such tag. He thus lets the information contained in romances circulate among his mental databases as architectural truth, corrupting his knowledge about the world that we assume has hitherto been relatively accurate. Among other literary characters belonging to this tradition are Arabella, the heroine of Charlotte Lennox's The Female Quixote (1752),1 who takes the fantastic events described in French romances as the accurate representation of reality, as well as the already mentioned Katerina Ivanovna from Dostoyevski's Crime and Punishment and Charles Kinbote from Nabokov's Pale Fire, both of whom wax delusional by failing to keep track of themselves as the sources of their fanciful representations about the world.
The category of such "Quixotic" protagonists can be further expanded if we consider characters whose source-monitoring is somewhat compromised, though not to the degree that renders them unquestionably mad, such as Richardson's Lovelace [Clarissa) and Nabokov's Humbert Humbert (Lolita).2 For a literary critic exploring fictional narrative's manipulation of our metarepresentational capacity, such characters as Lovelace and Humbert are particularly fascinating: not only do they conflate their visions of reality with the more "real" reality, but they also drag their readers along into that perceptual quagmire.
That's what such novels do then. If in Don Quixote and The Female Quixote, the failure to keep track of sources of certain types of representations was restricted to the title characters, making them the locus of madness, Clarissa and Lolita diffuse this fascinating failure among characters and readers, making us experience if no
t a bout of insanity then still an occasional feeling of mental vertigo. This feeling, captured in part by the literary-critical term unreliable narrator (more about this term later), is predicated upon our anxious (though not, of course, articulated in these terms) realization, as we read on, that we have been tricked by the narrative into losing track of sources of certain representations.
Consider Lolita and its first-person narrator, Humbert Humbert. We realize (so to speak) that going back and retroactively turning representations into metarepresentations by supplying source tags such as, "It was Humbert's idea that Lolita has been sexually interested in him, for in reality she has not," is a treacherous undertaking. What if she has been, a little bit? Whom can we trust now in figuring that out? Which source tags should we retain? Which discard? Which reweigh as to their relative truth-value? Having processed some representations as architectural truths within the world of the novel, are we now supposed to scrap the results of that processing? And if we do, where is the guarantee that our new assignment of truth-values will hold for the next fifty pages in this kind of story? Some writers never fully resolve the source-monitoring ambiguity cultivated by their narratives, leaving it to those readers who appreciate this kind of mental game to enjoy Lolita and Pale Fire; others may grudgingly settle on decidedly battered and compromised versions of the "real."
Of course a reader who hated Lolita yesterday because she felt that there is no stable ground in the story from which to judge the truth-value of any given episode (read: no reliable source of information; too much metarepresentational ambiguity) may start liking it tomorrow, influenced, for example, by her classmates' discussions of the novel. Circumstances change, minds change, readers change. But even as we reinvent ourselves every second, we still cannot help monitoring sources of our representations and constantly reweighing the relative truth-value of those represen
WhyWeReadFiction Page 10