Janet H. Murray
Herbert Simon's proposal of a theoretical basis for literary criticism is straightforward and-for an AI researcher-surprisingly modest. He recognizes that what the social scientist belittles as inconsistency the humanist cherishes as ambiguity. He is at pains to affirm the inherent complexities of texts.
Simon proposes bringing together the two disciplines of cognitive science and literary criticism in order to jointly explore how meaning is created in the human mind. His proposal rests on the assumption that there is a congruence between the structure of texts and the structure of minds. This is a familiar assumption from Chomskian linguistics, although here as there, it is easier to see how this congruence operates in syntactical systems than in semantic systems. Since literary texts are perhaps the most complexly structured semantic systems humans produce, Simon is proposing either an absurdly reductive or a breathtakingly heroic enterprise, depending on your degree of confidence in his methodology. In either case his is a mission of high seriousness and is in itself a welcome addition to the discourse of literary criticism, which lately seems too often possessed by an obscurantist, ironic, and self-referential cleverness that does not address the human condition so much as it displays a profound anxiety about it. As a Victorianist I am attracted by the simple earnestness of Simon's challenge.
Humanists may well be put off, however, by a fellow who wants to bring "precision" to the evocative, who speaks of "extracting" meaning and of the creation of ambiguity as "a major target" of the writer. This mechanistic model, blurring the differences between symbol systems in silicon chips and in flesh and blood is inherently repellent to the humanist who is interested not in quantifiable, nameable affects but in messy, mysterious passions. Yet as children of the enlightenment and as beneficiaries of Darwin and Freud, humanists have to be open to the claims of the rationalist and to resist the defensive posture. Instead we need to confront Simon's model of literary activity with two key questions: how far does it go in describing literary processes? how can we make Simon's insights useful?
How far does Simon's model go? The kind of structures that he proposes seem recognizable but operate at a lower level of organization than the structures of texts. Simon perhaps would not agree, but it seems to me that what he is describing is the micro level of processing, and what literary criticism tries to describe is the macro level.
Simon's persuasive description of "meaning," "evocation," and "context" all rest on the model of a series of associations being "activated" by a word or a text. The notion of multiple and overlapping associations, of ever shifting patterns of arousal, seem consistant with ordinary experience and helpful as a way of thinking about what used to be called "resonance" and what is now often called the "indeterminacy" of a work of art. But Simon's analysis offers only a picture of the lowest level of functioning. As he says, "[n]eurons or symbols, they are simply patterns." Yes, but what exactly are these patterns and how are they created and modified? Since it is the patterning that counts, the flatness of the arousal model does not offer us a helpful enough tool for looking at the creation of meaning.
Literary criticism starts from the other end of the scale. It describes the macro structures using a rich variety of representations from the rhetorical to the cultural. We can describe figures of language like metaphors and similes, larger formal structures like genres and subgenres, and most ambitiously we can tease out complex intersecting cultural structures of ideology, psychology, and even the dense world of "values." In short at the critical level there is a tremendous richness of available patterns but no shared notion of how these patterns are part of minds rather than texts.
Is there then any commonality between these two viewpoints? Is this a difference of granularity (like the granularity difference Simon intriguingly posits between non-representational and representational painting)? Are the activated neurons the pointillist dots that, seen correctly, make up the larger patterns of sonnet or Oedipus complex or patriarchal imagery? If so, what does it help to know this and how will we ever get to the interesting territory that lies between them?
This brings us to the question of usefulness. It seems to me that the most promising part of the cognitive model is its invitation to literary critics to think in terms of process rather than product. The great promise of building a persuasive cognitive model at the micro level is that it can perhaps serve as a model of how the larger patterns of cultural and formal elements come together and are changed by the artist or by the work of art. Jane Eyre is valued today because it rearranged the cultural codes by which meaning was "extracted" (to use Simon's term) or "constructed" (to use the current critical language) in its society. It is a masterpiece because of a cognitive breakthrough in imagining women's lives (among other things), in foregrounding desires and perceptions that society did not, until Jane Eyre was written, have patterns to convey and affirm. What is the process by which a writer creates meaning that transcends the conventional associational patterns of the society? Can an understanding of the process of formation of gestalts at the lower levels of cognitive processing offer us a procedural model with which to understand the higher processes of creativity?
Can we go from the notion of an activated set of neurons to an understanding of how those neurons are organized into patterns? Would this further the enterprise of literary criticism? Let us try some steps toward the middle. Let the cognitive scientists try to model the mind's extraction of meaning from simple literary structures-perhaps jokes. Simon mentions puns in his article. Why not start there? Can the cognitive scientists come up with a model of generation and understanding of puns that, like a Turing test, would "feel right" and not feel reductive to the ordinary humanist in the street? Meanwhile, from the other end of the bridge, let the literary critics turn to an examination of the processes of creation and of reading. I would argue that in successful creation, conflicts within the psyche, between the individual and the society, and between the emerging artist and the inherited tradition all line up like nested spheres, and the writer's efforts to form a new gestalt in any one of these spheres rearrange the forces in the other two. Can we model this process as carefully in such a way that it would be recognizable to those interested in seeing it not as psychological or political or aesthetic but as pure cognition?
Yet at the same time as I would welcome such an endeavor I would still be skeptical of cognitive science's imperialism (as Simon playfully calls it). In fact in the same spirit, I would propose a counter imperialism drawn from my teaching, in a course on interactive fiction, Colby's early work in modeling human psychological processes on the computer. These experiments are a good measure of the promise and limits of the cognitive approach. Colby's computer-based psyches-paranoid man and a woman with the repressed hatred for her father-are too wooden and predictable to be persuasive models of human psychodynamics. But as literary characters they are comically pleasing and plausible, just as any "flat" character can be plausible within its own medium. That is, they fail (in my opinon) as a form of scientific investigation but they succeed as a form of genre fiction. The literary critic may see all of cognitive science in a similar light: an underimagined and formulaic view of the human psyche, engaging and widely recognizable, but impoverished beside the work of a George Eliot or a Tolstoy.
As a humanist I have a core belief in the wedding of human cognition to our bodies and to our lived experience; I do not expect that our complex and richly textured emotional life will be captured by quantitative or mechanistic models. Yet I would still welcome the collaboration Simon proposes as a good way for literary critics to learn the extent to which their concepts can be made "precise" without reductiveness, and for cognitive scientists to test the limits of their very powerful forms of representation.