part 5
Herbert Simon
We should not suppose that it is always or usually the author's intent to be wholly consistent. Imperceptible inconsistency is supposed to be one of the finest marks of the political orator's art. (The inconsistency is easy, but with national media, the imperceptibility becomes harder to sustain.) Punning is another reminder that inconsistency may reflect intent rather than accident.
The term "ambiguous" sounds less pejorative than "inconsistent," so let's substitute it. In the kind of writing we call "literature," possibly in contrast with "scientific" writing, the creation of ambiguity rather than its elimination may be a major target of the writer's efforts. This is nowhere more evident than in Joycean and post-Joycean writing, but we should not think of it as peculiarly modern. Without trying to be exhaustive, let me propose several different kinds of ambiguity that are important to art.
In novels, we object to "flat" or "two-dimensional" characters. Such characters are too simple and, what is worse, predictable. In a great novel (i.e., The Remembrance of Things Past), we catch partial glimpses of the characters, as we do of real people. Over time, character changes, both of itself and because successive glimpses reveal new facets. And even on the last page, do we really know the people we have met in the book? Can we extrapolate the Baron Charlus of The Past Recaptured from his early appearance at Balbec? And which is the real Charlus?
Do we know our friends, our spouses? Do we know the town and the society we grew up in? Can they provide us with no surprises? We wish literature to reflect this fundamental and admirable characteristic of life--its ambiguity. Hence, the author is often absorbed in creating multiple meanings, craftily interwoven.
Ambiguity is as prominent in painting and music as it is in the literary arts. The banal example in music is the diminished seventh chord, which, with its four enharmonic equivalents can signal a dozen keys, allowing the composer to transpose without other warning from one to another. But the multiple interpretation of pattern in music goes far beyond this simple device and is especially endemic in such composers as Beethoven and Stravinsky.
In painting, ambiguity takes the most varied forms, from the calculated trickery of Escher and of the Cubists to what one might describe as the vagueness of much Abstract Expressionism. A portion of a canvas that retains no ambiguity, that allows only a single interpretation, usually appears dull and lifeless.
Notice that the function of ambiguity is not to drain the work of art of meaning but to enrich its meaning by suggesting that one interpretation does not exhaust it. Nor does this kind of ambiguity "leave the meaning to the reader." On the contrary, it is the task of the author to write in such a way that a multiplicity of meanings can be teased out by the reader. They are not to be created from whole cloth but are hidden in the work of art in order that they may be discovered. Of course the author cannot prevent the reader from finding additional meanings (or missing those intended to be found), but few authors regard themselves as simply creating Rorschach inkblots to evoke wholly personal patterns from readers' memories.
At the other extreme, we tend to denigrate art, visual or literary, that evokes pre-existing memory structures in too simplistic a way. Thus Norman Rockwell depends for his effects upon memory structures ("Mom and apple pie") stored reliably in the heads of his viewers and triggered into awareness by very literal and explicit clues.
The art of "high culture" distinguishes itself from that of popular culture in calling for more subtle and elaborate sequences of evocation and association or (more snobbishly) for evocation of relatively esoteric structures that are stored only in properly educated memories--memories that have had extensive exposure to the appropriate Canon. New Yorker cartoons provide an adequate if frivolous example. Only those who are in touch with particular aspects of the New York scene will be likely to understand them.
Art is called representational when it denotes something outside itself, nonrepresentational otherwise. Until rather recent times, virtually all art, as distinguished from design, was assumed to be representational: Art was supposed to imitate Nature. The major exception was music; in absolute music, which encompassed the bulk of classical music other than song and opera, the only "meaning" was supposed to be the set of tonal relations placed there by the composer and detected by the listener. The meaning of music was the musical pattern, rhythmic, tonal, and harmonic. It would, of course, evoke emotion, but it had no other reference to anything non-musical.
Only in our own century have we had "absolute" literature or "absolute" painting that undertake to emphasize internal relations as their meanings and eschew reference to things outside. Examples, even approximate examples, are hardest to come by in literature. We might think of The Waste Land as nonrepresentational, or Finnegans Wake, but in no such strict sense as a Bach Fugue or a Mozart Sonata.[3] "Enigmatic," in that the denotations are obscure and often kaleidoscopic, might be a better term than "nonrepresentational" for such examples. Poetry, with its attention to phonetic relations, has generally a larger explicit nonrepresentational component than prose.
It is a matter of terminological preference whether we want to use the word "meaning" broadly enough to encompass the nonrepresentational components of pattern. Apart from the question of terminology, there are no particular problems in seeking out such patterns in a work of art, whether it be music, painting, or literature. It is really the representational component in painting and literature, rather than the nonrepresentational component, that is problematic.
What does a Renaissance Venus and Adonis represent? Most of us consider Venus and Adonis to be mythological figures who never trod the Earth. How can a painter represent them? What one sees on the canvas is a figure that is recognizable as a beautiful woman and another recognizable as a handsome man. From their stances in relation to each other, one can perhaps infer some motives, hopes, expectations, and intents. (Or one can judge consistency of those stances with motives, hopes, expectations and intents in the myth they are supposed to portray.) But what do the figures represent? At most, the models who posed for them; at least, a generic woman and a generic man.
What is represented, then, in this kind of "representational" painting is not some actual scene or event but a hypothetical scene assembled from components that could actually be seen. A representational painting is like a molecule synthesized by chemical ingenuity (i.e., nylon or aspirin). Nothing like it exists in nature, but the atoms of which it is composed are real enough, as are the chemical forces that bind them together.
Of course, by this standard it could be argued that a Kandinsky painting represents a scene that could be seen through a microscope. The difference is merely one of grain size. If the grains are small (as in Kandinsky or Pollock), comparable to what we see under the microscope, we call it nonrepresentational. If the grains are large (as in "Venus and Adonis"), whole human figures in a possible landscape, we call it representational. Similarly, in Joyce and Eliot, the represented elements are small; in Dickens or Flaubert, they are much larger--but still arranged in contrived patterns that correspond to only fictional events.
The notion of "representational," then, makes some appeal to the notion of "possible." Something is representational to the extent that it denotes real things in mutual relations that could possibly be real, even if we know that in fact they are not. Madame Bovary does not violate any of our beliefs about what 19th Century small-town bourgeois French society was or could be like or about how people in such a society did behave or could have behaved.
In the same way, a physicist could describe a star like our Sun, but only half as massive, with planets about it, but at different distances from it than our planets are from our Sun. The movements of this solar system could be represented, although the objects it denotes presumably do not exist. Such a description would be representational in the same sense that a novel is representational.
What is possible depends on which laws of nature are relaxed and which are enforced. Science fiction is a peculiar genre in this respect, since it opens the doors of possibility (at least physical possibility) more widely than most other forms of fiction do. If a novel depicted people who were so monstrous as to be unbelievable (impossible?), would we regard it as representational? As science fiction? Or just as a bad novel, like the novels whose characters are cardboard? In representational literature, we seem to insist more strictly upon the laws of human nature (however imperfectly we understand them) than upon the laws and boundary conditions of physical and biological nature.
The writer of a piece of representational prose describes a possible world, and we grasp the meanings to the extent that we are able to evoke from the words and our memory a world like the one described. Since we carry around in memory schemas or scripts for a great variety of possible elements that can be combined into scenes and narratives, writers draw on knowledge of the nature of these schemas to predict what meanings, what possible worlds, readers will extract from their words.
My discussion of ambiguity suggests that there may be no such gulf between the meanings of the author and the meanings of the text as is presumed in some contemporary theories of criticism. The meanings of the author can be characterized by what he or she had in mind: what was evoked from the author's mind, what was actualized while writing the text. But in many, perhaps most, cases authors will in fact intend to evoke similar meanings from their readers, especially when reader and author belong to the same culture. To the extent that they succeed, the meanings of the text will be the same as the meanings intended by the author. We might even assert that the craftier the author (the greater the mastery of the craft) the closer will be the correspondence between the author's meanings and the meanings that critics or readers can find in the text.
Now the fat is on the fire: I have mentioned critics. I find it impossible to interpret "the meaning of the text" in any other sense than "the meaning of the text to X," where X is some person or computer capable of having intentions. If the meaning of the text is not the meaning to "lay" readers, it must be the meaning either to a critic or (to borrow the word "representative" from the economists) the meaning to a representative member of the language community.
One responsibility of critics is to identify the meanings that can be evoked from texts--both the meanings intended by the authors and the meanings that, intended or not, are consistent with the ambiguities. And for the Great Texts--the Bible, Shakespeare, Homer, and the others--the ambiguities are inexhaustible, a permanent lode of treasure for scholars.
I tint this last statement with only the palest wash of irony. The meanings that the Great Texts have for us cannot be separated from the generations of scholarship that have gone into interpreting them, and it becomes almost irrelevant whether those meanings were "really" in the texts when they were written. It is as irrelevant as whether the rule of the Miranda case is really in the Bill of Rights of the Constitution. For a person who shares the appropriate cultural heritage (has the right schemas in memory), the Texts can evoke the future as readily as the past: Homer can evoke Joyce just as Joyce can evoke Homer.
Scholarship attaches new meaning to the question of "who was the author of Shakespeare's plays?" Shakespeare must now share that authorship with all those who have commented on him, borrowed from him, plagiarized him, been compared with him, distanced themselves from him. When we read Shakespeare, any or all of the comments of these co-authors may be evoked as part of his meaning--or perhaps I should say "as part of the meaning of his text."
I must pause here to comment briefly on an issue that bulks large in contemporary talk about liberal education. It is not hard to translate into the terms we have been using here the battle that is going on between the defenders of the Canon, the core of Great Books central to the traditional curriculum, and the opponents of ethnocentrism.
The defenders of the Canon argue, correctly, that only an informed mind can read deeply and extract rich meanings--even those implanted in the text by the author, to say nothing of those added by the text's subsequent role in the culture. The opponents of ethnocentrism argue correctly that new meanings may be evoked from a text by a reader who has stored in memory knowledge of other cultures and are unlikely to be evoked by a reader who hasn't.
Juxtaposing these two correct arguments we are led inexorably and unhappily to the conclusion that life is full of dilemmas and hard choices. Since we have not time to read all the texts, we must devise some scheme of sampling. And different schemes will assign different weights to the old and the new Canons.
We should like to select texts that will enhance our abilities to understand, and even empathize with, cultures other than our own and, at the same time, texts that will enable us to hold common discourse among ourselves. Adherence to a Canon is surely conducive to ethnocentrism; sampling widely and variously is inimical to common discourse of any depth, for each text must be examined in the absence of much of its context (its meaning).
The structure of this dispute is identical with the structure of the common dispute between breadth and depth; and it should probably be settled in the same way. An inverted "T" is an attractive shape for an education. The broad base assures breadth; the tall projection assures depth (and in our case, an area of common discourse). Constructing both the base and the riser of the T calls for sampling, but the exact content of the sample does not much matter.
For the broad base of the T, we can go to anthropology and some kinds of sociology, to history, and to world literature (in translation, alas). To give our T depth, we can use the tried and true "making of the modern world" formula, the history of Western civilization and/or some shelf of "great books." The history, of course, should be mainly social history rather than the old-fashioned kind.
The University of Chicago in the 1930s (educational philosophy so often turns out to be nostalgia for the undergraduate curriculum that the philosopher experienced) used its Humanities and Social Science survey courses to provide just such a broadening and deepening experience. One who had this training was able to bring to bear upon texts from the Western tradition (i.e., most of the texts he or she was likely to encounter) a rich context to illuminate their meaning. This context, in fact, incorporated a wide range of views of class and gender. At the same time, the base of the inverted T exposed the assumptions implicit in this Western tradition to sharp examination from the vantage points of other cultures.
Other formulas can produce much the same result. At Columbia, it was a particular selection of great books; at the University of Wisconsin it was Mickeljohn's experimental college built around Henry Adams's Mont-Saint-Michel and Chartres and his The Education of Henry Adams. The exact formula doesn't matter. What does matter is the opportunity to acquire knowledge that will assure the evocation of rich meanings from any text, with a particular focus on meanings that relate to the culture in which one is likely to spend most of his or her life, but with the external view, also, of that culture that is provided by exposure to alternative Canons.
To develop these educational issues at length would call for another paper as long as this one, a paper that would be out of place here. My purpose in introducing the issue at all is to show how the theory of meaning I have been explicating can help us examine questions like these. The theory provides a framework within which reason can be applied to the choice of texts to be sampled.
Perhaps it is the parasitism of scholarship upon literature that creates the impression that there is a disembodied text (independent of author or particular readers) which is to be interpreted. That text is the possession of the language community--more strictly of the community that know its scholarly context. Only in them can all of the text's richness of meaning--in this extended sense--be evoked.
For any text that is preserved over years and centuries, the ownership becomes even more ambiguous. Now there is not only the question of who interprets it but in which historical context, past or present, the interpretation is carried out. We can read the Bible (or Homer, or Chaucer) in the context of the author's culture or our own--or in the context of the ideas of medieval Europe or 16th century China.
Whether authors would be gladdened by this kind of extension of their meanings is another question. We could guess at the answers. My guess is that Stendhal would be rather appalled,[4] Joyce entranced, and Proust bemused. Tolstoy, who conceived himself as writing down truths, would be indignant: it is his meaning that he would wish to have preserved, nor would he want any reader to actualize a different meaning. But now I am playing the critic's game of inventing meanings for authors and should leave this subject.
In sum, attention to "the meaning of the text," more or less distinct from "the author's meaning," or "the meaning to the reader," has produced a large part of that luxurious jungle growth we call literary criticism. There is no reason for the Schools to quarrel--they are simply using different reading processes and employing different contexts in which to read their texts. Labeled for what they are, and stripped of claims of exclusiveness or priority, any of these processes and contexts may define interesting, or even valuable, activities. And there might be much merit in aiding undergraduates to develop skill in several or all of them.
I have already said a good deal about readers, since I have been at pains to insist that critics and authors are readers. Other readers can, of course, read like critics, searching for authors' meanings, exploiting ambiguities, drawing upon the cultural riches that surround the Great Texts, or perhaps inventing other games of their own. Readers have even made counts of word frequencies in Joyce's Ulysses, in Bunyan's Pilgrim's Progress, and in the Federalist Papers that are not without interest.
Perhaps a word should be said about (or in behalf of) the reader who entrusts himself or herself to the author and only requests that interesting meanings be evoked by what the author has written. Such a reader is sometimes called "casual" or "escapist," but it is possible to suppose that it is precisely this reader whom most authors have had in mind.
Further, it is possible that many or most readers who belong to the same culture may detect approximately the same meanings in an author belonging to that culture. In an earlier and more innocent age of criticism (and authorship) this may even have been the normal situation: an author recording meanings evoked from his or her mind that were, in turn, evoked in the minds of readers who perused the author's text. What a pleasant, simple world for authors and readers, if perhaps a somewhat impoverished one for those who earn their livings as critics!
In a slightly more piquant situation, the reader wishes to penetrate another culture: an American reading the English translation of Kawabata's Snow Country, say. (A geisha becomes angry when her lover calls her "a good woman" ("yoi onna"). But is Japanese "anger" American "anger"? And is "yoi onna" a "good woman"?) Here, one would almost call it a cheat if the reader simply allowed the text to evoke meanings steeped in American cultural assumptions.
But what is the source of leverage to avoid that evasion? Are there human absolutes, unchangeable from one culture to another, to provide the fulcrum that allows the reader to pry out and interpret the subtle differences? There must be some such invariance derivable from the text if the meaning is to be extractable even in part. But now I have moved into the theory of translation, which I cannot undertake to explore further here.
Meaning for the reader is, at bottom, no different from meaning for the critic or the author, although a given text may correspond to quite different meanings for all three. When I say "no different," I refer to the meaning of "meaning." The meaning we inject into a text or derive from it consists of all of those cranial symbol structures that are evoked by its making or reading, including those intensional tests that map part or all of the reading on outside-world denotations.
When I visualized the lift at the Grand Hotel in Balbec that Proust described for me, I saw (faintly and vaguely, in my mind's eye) an elaborately decorated open cage that rose from the lobby, its distinguished passengers visibly levitating to their rooms above. When later, on a visit to Balbec (Cabourg), I actually saw that lift, its meaning--the picture evoked from the retinal image--corresponded almost exactly with the mental picture that Proust's words had evoked in me. That afternoon, sitting on the beach, I drew a sketch, a quite literal sketch, of the hotel itself. Whether it came from my memory of Proust's words or from the scene before my eyes I can hardly say.
In these pages, I have given a cognitive science interpretation of the enterprise of literary criticism, showing how the critical enterprise can be interpreted in terms of the evocation of texts from meanings and of meanings by texts. Armed with such an interpretation, we can see what is at stake in an investigation of the sources of imagery in Coleridge's Kubla Khan or the mnemonic devices of the Homeric epics and the songs of Yugoslav bards.
But what possible utility can such a translation have for humanists? Let me, without attempting either a taxonomy or an exhaustive listing, provide some examples of ways in which it might be profitable to think of criticism in the language of cognitive science, including computer simulation.
Since my first example is not literary at all, it depends on analogy for its relevance. I have already mentioned that Gordon Novak, now at the University of Texas, wrote an interesting computer program, which he called ISAAC, that reads problems at the end of the chapter of a physics textbook and converts the information conveyed in the problem descriptions into representations stored in the computer's memory. These representations are so "pictorial" that ISAAC can use a simple program to draw pictures of them on the computer terminal screen. Having drawn such a picture, ISAAC can also reinterpret the internal representation with equations describing the physical forces that are at work and can solve these equations.
Here, then, is a computer program that can translate from language to mental pictures and from mental pictures both to real (visible) pictures and to mathematical equations. Let us pass over such philosophical issues as whether ISAAC understands what it is doing. (I think it does; John Searle thinks it doesn't. But no matter.) What ISAAC clearly does do is to give us a tangible idea of what is involved in converting meanings among natural language, pictorial, and formal symbolic representations. It tells us, for example, what it has to know (to have stored in memory) in order to bring the trick off, and it exemplifies in detail the processes that can be used to do it.
Since such translations (at least between language and pictures) are central to much literary construction, the insight provided by ISAAC should have some interest and value for us here. The important thing is not that a computer is doing the job, but that we gain from its performance a better understanding of what the job is.
Let me take one step closer to literary language. Linguistics began its work with the sentence (or even the isolated word) and has gradually been working its way up to paragraphs, passages, and occasionally whole stories. The task of constructing "story grammars"--accounts of the structures of tales and the processes that understand the tales by discovering these structures--has received attention both from linguists (e.g., Gümlich and Raible, 1977) and from cognitive psychologists (e.g., Mandler, 1978), as well as from bridging figures (e.g., de Beaugrande, 1980). Considerable success has been attained in analysing stories at the level of complexity of, say, Little Red Riding Hood. Story grammars begin to tell us how such components of meaning as plot, motivation, and character are conveyed.
The work that I have mentioned in these paragraphs--both spanning from literature to its psychological foundations and from cognitive science back to literature--provides some prototypes for the kinds of inquiry that can in time create a broad and well-traveled bridge between the two cultures.
There is no great mystery in the meaning of "meaning."
If we agree that meanings are held in heads and that heads contain complex networks of neurons, then we can locate meanings in the symbol structures that these networks contain. Neurons or symbols, they are simply patterns. And the patterns have meanings because they can refer (point) to each other and especially because some of them can denote (through perceptual tests of the kind I described earlier) things, relations, and events outside the head.
The theory of literary criticism has to do with the ways in which meanings are used to generate text in natural languages and the ways in which the perusal of texts evokes meanings. Viewed in this way, criticism can be viewed (imperialistically) simply as a branch of cognitive science. I have tried to sketch out some of the rich modes in which criticism can be pursued and has been pursued historically. A closer tie of criticism with cognitive science will surely lead to the invention of new modes, which may turn out to be as fascinating and valuable as those already pursued.
I have talked of meanings and criticism descriptively rather than normatively. An outsider, viewing the internecine battles that seem to go on constantly within the commmunity of critics and theorists of criticism, wonders why they cannot be settled easily and pacifically. Each School appears to describe some particular mode of evocation, hence of meaning, and then to claim it as the correct one. If the claims of uniqueness and exclusive correctness were abandoned, as they surely must be, peaceful coexistence could be wholly restored. Of course that might not be as much fun as the current noisy combat.
I have spoken mainly about literature, a genre in which the transmission of true declaratory information is at most one of an author's intents and often not the most important. There is, of course, also criticism applied to expository prose, where we direct our main attention to veridicality, clarity, absence of ambiguity, and simplicity (perhaps in that order). This does not mean that we must exclude literary qualities from expository prose, although they may have to compete with the qualities proper to the genre. Certainly some of the Great Texts in science have important literary as well as expository virtues.
In writing this essay, I have intended to write expository prose. In particular, my main intent was to set forth a clear theory of meaning and to apply it to the enterprise of literary criticism. I tried to avoid technicality in doing this, and if I have sacrificed clarity, I will refer the reader to some of my more technical writings on these topics, which may throw light on what is dark here. I recommend in particular Chapters 3 and 4 of The Sciences of the Artificial (1981) and my recent essay on The Information Processing Explanation of Gestalt Phenomena (1986).
My own position here, while the position of an outsider, can be interpreted as proposing yet another school of criticism. Using the dichotomy proposed by Professor Stephen Greenblatt between resonance (that is, knowledge) and wonder, my theory would seem to fall solidly on the side of resonance. I have been exploring the contexts from which meanings arise; I have not been expressing the wonder that meanings can evoke.
But scientific analysis is never very distant from wonder. The great Dutch mechanician, Simon Stevinus, took for the motto on his crest: "Wonder en is gheen wonder"--wonderful, yet not unfathomable. A snowflake is still wonderful, perhaps more wonderful, after we have discovered the fractal origins of its design. Meanings are still wonderful after we have traced out the cognitive processes that underlie them. Science adds to the wonders of appearance the wonders of explanation. It follows the path from wonder to resonance, and then back to a deeper wonder.
At the outset, I gave my principal reason for addressing this topic: it seemed a good site for a bridge between the two cultures. If the span I have erected seems flimsy or unaesthetic, I hope that others, when dismantling it, will replace it with a better one.
Professional competence in a domain of the humanities, like competence in a domain of science, requires the accumulation of a great deal of specialized knowledge. We cannot expect to master the content of more than a very few domains in any great depth. What we can hope to do is to work toward a common understanding of the mental processes that all of us use to extract meanings. However distinct and dissimilar the domains, our minds, fashioned from the same raw stuff and employing the same basic symbolic processes, must have a great deal in common that we can share.
If scientists and humanists have only this small common ground, it is very important ground. They can share the common experiences of being human. They can share a concern for the polity and the society in which they carry on their lives (drawing the boundaries as wide as the whole world). And they can share the processes that their minds use to deal with the inner and outer domains, personal, professional, and civic, in which they are immersed--the processes that evoke and manipulate meanings, from texts and from the world.
This paper had its origins in one of the Hitchcock Lectures that I gave at the University of California, Berkeley, in 1990, but has since undergone extensive revision. A number of my colleagues, including David Carrier, John R. Hayes, Alan Kennedy, Erwin Steinberg, and Gary Waller have provided me with extensive and insightful comments on earlier drafts. I thank them warmly for their help while pronouncing the usual absolution and taking full reponsibility on my own shoulders for the final product. I am grateful also for the insights I obtained from those who pioneered in the directions toward which this essay points. At the risk of omitting others, I should like to mention in particular Robert de Beaugrande, Walter Kintsch, and Roger Schank.
3. And The Waste Land has become decidedly more representational as we have learned about its relation to the life of Eliot.
4. Stendhal claimed to write for a specific culture, not the one he endured, and to live in confidence that its time would come.