Online Magazine of the Visual Narrative - ISSN 1780-678X
Issue 1. Cognitive Narratology
Narratology as a cognitive science
Author: David Herman
Abstract (E): Using the 1999 MIT Encyclopedia of the Cognitive Sciences (= MITECS) as a springboard for discussion, this essay sketches out some of the implications of recent work in cognitive science for narrative theory. Inversely, the essay explores how current modes of narrative-theoretical inquiry bear on research in the cognitive sciences, which encompass such fields as cognitive anthropology, computational intelligence, linguistics, philosophy of mind, and cognitive psychology. MITECS lacks entries for NARRATIVE or STORY, and more generally the text omits discussion of many narrative-pertinent concepts that promise to throw light on key research questions in cognitive science. My essay thus uses MITECS as a vehicle for opening new lines of communication between areas of research that stand to gain from being brought into a closer, more synergistic relation with one another. More than this, the essay makes the stronger claim that narrative theory should be viewed as a subdomain of the cognitive sciences.
Abstract (F): A travers un compte rendu du récent "MITECS" (=MIT Encyclopedia of the Cognitive Sciences, 1999) le présent article discute les enjeux des recherches cognitives contemporaines pour la théorie narratologique. Corollairement, il entend démontrer l'impact des développements modernes en narratologie devraient intéresser plus d'un champ cognitif, comme par exemple l'anthrolopologie cognitive, l'intelligence artificielle APO, la lingusitique, la philosophie de l'esprit humain ou encore la psychologie cognitive. MITECS n'a pas d'entrée particulière pour des concepts comme NARRATION ou RECIT et très souvent le volume omet carrément de prendre en compte un certain nombre de concepts narratologiques susceptibles d'éclairer des questions clé dans les recherches congitives. L'article souhaite indiquer quelques points et problèmes communs à la narratologie et à la recherche congitive, et contient un plaidoyer pour un rapprochement raisonné des deux disciplines. De manière plus fondamentale, il tente de démontrer que la narratologie doit être considérée comme un secteur particulier des sciences cognitives.
Keywords: cognitive narratology
The purpose of this essay is threefold. One of my aims is to sketch out some of the implications of recent work in cognitive science for narrative theory. A second aim is to consider how, inversely, current modes of narrative-theoretical inquiry bear on the field of cognitive science. To be sure, the sheer scope and complexity of the issues involved would make it difficult to accomplish either of the two goals just mentioned, let alone both of them concurrently. Yet-and here I come to my third overall aim-The MIT Encyclopedia of the Cognitive Sciences, published in both a book and CD-ROM format in 1999, provides an indispensable guide in this context. (The page citations contained in this essay refer to the book version of the Encyclopedia. I did not have the opportunity to use and evaluate the CD-ROM version.) Theorists interested in exploring the cognitive dimensions of narrative can use the Encyclopedia-hereafter referred to as MITECS-as a kind of legend or key; i.e., MITECS articles afford entry-points into paths of inquiry that contemporary narrative analysts would do well to pursue more assiduously than they have up to now. Strikingly, however, MITECS itself lacks entries for NARRATIVE or STORY, and more generally the text omits discussion of many narrative-pertinent concepts that promise to throw light on key research questions in cognitive science. In short, both the organization of MITECS and the current state of narrative-theoretical research reveal the need for more dialogue between scholars working in the field of cognitive science and those engaged in the study of stories. The third aim of this essay is thus to use MITECS as a vehicle for opening new lines of communication between two important areas of research, both of which stand to gain from being brought into a closer, more synergistic relation with one another.
In what follows, I first characterize MITECS in greater detail and focus on some representative entries that suggest new possibilities for a cognitive approach to narrative analysis. Along the way I discuss developments in narrative theory that, though they seem not to have propagated themselves far enough into neighboring disciplines to affect the composition of MITECS articles, nonetheless have the potential to reshape central questions in the cognitive sciences. I conclude with some metatheoretical remarks about narratology-about what kind of theory the theory of narrative should be. Whereas narratology began as a subfield within structuralist literary and cultural theory, and looked initially to linguistics as its "pilot science," I suggest that both narrative theory and linguistics should instead be construed as resources for cognitive science. Or rather, narratology, like linguistics, can be recharacterized as a subdomain of cognitive-scientific research. From this perspective, both language generally and narrative specifically can be viewed as tool-systems for building mental models of the world. MITECS entries provide, in effect, a partial inventory of the contents of these two toolkits for model-building. Subsequent researchers can use the volume to work towards a more exhaustive, more detailed inventory and description of the representational tools at issue.
One final introductory comment about the formatting used in the present essay. Words in SMALL CAPS indicate terms for which there are entries-or, in the case of narrative and story, terms for which there might have been entries but were not-in MITECS. I use small caps only for the intial occurrence of each such term. Readers interested in extending the scope of this exploratory discussion, or developing particular aspects of it, can use the entries so marked as a basis for further investigation.
MITECS, Cognitive Science, and Narrative Theory
A truly landmark referrence work whose general editors are Robert A. Wilson and Frank C. Keil, MITECS includes 471 articles pertaining to six confederated disciplines that can be grouped under the umbrella field of cognitive science-philosophy; psychology; the neurosciences; computational intelligence; linguistics and language; and culture, cognition, and evolution. The hallmark of every article, most of them between 1,000 and 1,500 words in length, is a compactness and concision that should in no way be confused with sketchiness or superficiality. Article headings range from ANIMAL NAVIGATION, to CAUSAL REASONING, to DEPTH PERCEPTION, to INFANT COGNITION, to METAPHOR AND CULTURE, to PROBABILISTIC REASONING, to SMELL, to WORKING MEMORY. Each entry, furthermore, includes a list of references that amounts to a sort of "starter kit" for anyone wishing to gain greater familiarity with the issues discussed. One or more advisory editor oversaw the entries pertaining to each of these six subfields; advisory editors also wrote the useful introductory essays that open the volume (xv - cxxxii) and outline some of the most salient research concerns in the six subdomains. Robert A. Wilson supervised the philosophy entries; Keith J. Holyoak, the psychology articles; Thomas D. Albright and Helen J. Neville, the entries related to the neurosciences; Michael I. Jordan and Stuart Russell, the articles concerned with computational intelligence; Gennaro Chierchia, the entries on linguistics and language; and Dan Sperber and Lawrence Hirschfeld, the articles exploring links between culture, cognition, and evolution.
Here I can sketch out only a few of the ways in which this 964-page volume might assist narrative theorists working to develop a cognitive approach to study of stories. More precisely, I focus on aspects of several of the introductory essays and on a few sample entries, highlighting elements of MITECS that seem to me particularly consequential for ongoing narrative-theoretical research-as well as elements that seem particularly susceptible to being reframed by work in narrative theory.
MITECS's Introductory Essays
In his introductory essay on philosophy's contributions to cognitive science and vice versa, Robert A. Wilson notes that the problem of other minds has been a perennial concern among philosophers exploring the mind-body problem (xvii - xviii). Until recently, theorists have commonly drawn a distinction between first-person attributions of mental states and dispositions to oneself and third-person attributions of mental states and dispositions to others. As Wilson notes, however, research on INTROSPECTION and SELF-KNOWLEDGE has called into question how directly people know their own mental states. At the same time, inquiry into the THEORY OF MIND (or FOLK PSYCHOLOGY), ANIMAL COMMUNICATION, and SOCIAL PLAY BEHAVIOR has led to a reassessment of third-person attributions of mental properties to others. Specifically, empirical studies indicate that third-person knowledge is not as limited as was once thought, with attributions of mental properties to others being one of the basic strategies by which human beings organize their understanding of the social behavior of their conspecifics. If you come at me with your fist raised, I use my naïve yet quite serviceable theory of mind to impute to you anger and aggression as I figure out what to do next (e.g., run away).
For their part, narrative theorists interested in the cognitive implications of homodiegetic as opposed to heterodiegetic narration can benefit from a greater familiarity with the issues involved in this ongoing debate about the problem of other minds. Heterodiegetic, authorial, or "third-person" narration entails attributing mental properties to others, whereas homodiegetic or first-person narration seems to afford a discourse environment most favorable for introspection. As already indicated, however, the distinction between self- and other-oriented attributions is not as clear-cut as was once thought. Hence a narrator-protagonist like Daniel Defoe's Moll Flanders (Defoe 1981) spends as much time attributing mental states and dispositions to acquaintances, lovers, spouses, shopkeepers, constables, and fellow thieves as she does recording her own reactions, anticipations, emotions, and beliefs.
By the same token, researchers interested in the philosophical question of other minds would do well to consider narrative-theoretical accounts of the epistemological issues raised by the problem of heterodiegetic narrators' access to fictional minds (see, for example, Ann Banfield 1982, Dorrit Cohn 1978, Monika Fludernik 1993, and Alan Palmer forthcoming). It may be that people's everyday, folk-psychological attributions of reasons and motives to social actors involve processes radically similar to those at work in narratorial attributions of reasons and motives to participants in narrated worlds. If this is indeed the case, then folk psychology can be said to rely crucially on a style of thinking fundamentally narrative in nature. Everyday thinking about other people's thinking, especially their SOCIAL COGNITION, will be tantamount to situating cohorts in webs of narrative, in which they can be viewed as characters who are performing actions designed to circumvent or eliminate conflicts, maximize cooperation with allies, achieve desired goals, and so on. In short, when imputing mental states and dispositions to others, we may be recruiting from basic story templates identified early on by Tzvetan Todorov (1968), who viewed narrative as an instrument for negotiating the passage from an initial state of equilibrium, through an action or event disturbing that initial state, to a terminal state in which equilibrium is reestablished but on other grounds (cf. Kafalenos 1997).
Similarly, Keith J. Holyoak's introductory essay on psychology, which Holyoak defines as "the science that investigates the representation and processing of information by complex organisms" (xl), reveals how psychological research can both reshape and be reshaped by work in narrative theory. The essay provides a capsule history of the discipline of psychology (xli-xlv), as well as a synoptic account of the science of information processing (xlv-xlviii). Both of these sections suggest strategies for bringing narratological research into a fuller, more productive dialogue with developments in cognitive psychology. In the historical section, for instance, Holyoak discusses several investigative foci that have emerged as central concerns of psychological research, including what William James (1890) viewed as the hallmark of an intelligent being, namely PROBLEM SOLVING, i.e., selecting actions that will achieve goals; the internal models of spatially-structured environments that Edward Tolman (1948) characterized as COGNITIVE MAPS; and the SCHEMATA that Sir Frederic Bartlett (1932) viewed as mental representations that capture "systematic structural relations in categories of experience" (xliii). Bartlett's schemata would later be subdivided into dynamic and static knowledge representations-that is, stereotyped sequences of events ("scripts") and stereotypic states of affairs or situations ("frames") (Minsky 1975; Schank and Abelson 1977). Holyoak also mentions the pathbreaking work of LEV VYGOTSKY, who stressed the sociointeractional and cultural roots of human intelligence (Vygotsky 1978; cf. Frawley 1997), as well as GESTALT PSYCHOLOGY's focus on relational networks-e.g., the network of notes in a melody that can be played in different keys yet nonetheless recognized as constituting the same melody. At stake in such relations is the so-called BINDING PROBLEM, which centers around the way the components of a structure fill relational roles in a systematically organized way. As Holyoak notes, research on this problem informs modern-day psychological studies of such topics as ANALOGY and SIMILARITY.
Once again, these areas of concern in cognitive science can be linked to emergent topics of narratological research. Marie-Laure Ryan (1991), for example, has developed a model of plot in which the problem-solving activities of narrative participants is paramount. More specifically, participants can be seen as using a problem-solving criterion for action selection; i.e., participants act in ways designed to eliminate conflicts between wished-for possible worlds and the actual worlds (more precisely, "text actual worlds") in which it can be more or less difficult to realize the optimal or at least hoped-for states of affairs. Further, although early narratologists such as Roland Barthes (1971a, 1974) and Claude Bremond (1973, 1980) seemed unaware of analogous ideas being developed around the same time in the field of artificial intelligence (AI), Barthes proposed that readers make sense of stories by drawing on a "proairetic code" based on stereotypic knowledge of human actions, a "patrimonial hoard of human experiences" (1974: 204), and Bremond similarly grounded his "logic of narrative possibilities" in a "tableau of model sequences [rooted in] human behavior patterns acted out or undergone" (1980: 406). In other words, structuralist narrative theorists appealed to schematic world knowledge of the sort hypothesized by Bartlett and then recast in the form of scripts and frames by early AI researchers, themselves working on the problem of story comprehension (see Herman 1997 for more details).
Beyond this, however, recent research on narrative has yielded insights into some of the constitutive concerns of psychology outlined by Holyoak. For instance, Jerome Bruner, who construes narrative as one of the two main strategies for cognizing the world (the other being paradigmatic or logico-classificatory thinking [Bruner 1986]), has also argued that narrative enables the ongoing construction of social realities, not just post-hoc representations of them (Bruner 1991). In the context of a broadly Vygotskyean approach, Bruner's work suggests that stories should be examined as a fundamental sociointeractional resource for the development and refinement of human intelligence. The self may become what it is by virtue of its relation to the larger social order; but narrative may be one of the chief factors enabling the very possibility of this self-society relation and, a fortiori, intelligence as we know it. Sociolinguistic research by Elinor Ochs et al. (1992) on dinnertime storytelling as a theory-building activity, a resource for perspective-taking, and a vehicle for family-based processes of children's socialization supports the claim that Vygotsky-inspired psychology should focus on narrative as a primary object of inquiry.
Further, Mark Turner's (1996) work on a principle that he terms parabolic projection-or narrative imagining-suggests fundamental interconnections between stories and reasoning by analogy. Turner argues that, just as tellers of parables intend for recipients to "draw a lesson" from the stories they tell by applying those narratives to situations of which they can be interpreted as analogues, human beings in general work to make sense of novel experiences in terms of source stories that they project upon more or less difficult-to-process circumstances and events. From this perspective, it was perhaps inevitable that the concept "death" was personified in works ranging from Euripedes's Alcestis to Chaucer's Pardoner's Tale to the Holy Sonnets of John Donne. Recast as something brought on by an agent who imperiously removes our fellow mortals from this life, the atelic process of dying could be conceptualized as a bounded sequence of goal-oriented-and eminently tellable-actions in which those who have died are defeated by an antagonist replete with malice aforethought. More generally, narrative is an important resource for analogical thinking; it provides a cognitive and communicative environment in which dynamic, point-by-point comparisons can be drawn (or implied) between ostensibly discrepant situations and events. At the very least, the mutually enriching relation between narrative and analogy-with stories supporting analogical reasoning and analogies undergirding the situated design and interpretation of narratives-deserves further study.
When juxtaposed with recent work in narrative theory, Michael I. Jordan's and Stuart Russell's introductory essay on "Computational Intelligence" suggests still other mutually enriching relations and additional opportunities for studying them. The authors chose this title over, for example, "Artificial Intelligence," because theirs is broader in scope and encompasses what have emerged as two different approaches to "the problem of understanding intelligence in computational terms" (lxxiii). One approach treats the problem as an engineering goal, i.e., an effort to create intelligent machines; the other approach treats the problem as an empirical science whose aim is to develop computational models of human intelligence. (As Jordan and Russell point out, in practice if not in principle these two approaches often overlap.) In both approaches, developing computational models for generating and comprehending narratives can be viewed as an important (indeed, crucial) area of endeavor-one with implications for many of the issues associated with research on computational intelligence generally.
For one thing, a system that can produce and/or understand even moderately complicated stories-whether or not the system achieves this in a way that replicates human narrative competence-will necessarily be based on what can be called an INTELLIGENT AGENT ARCHITECTURE. "Such an architecture," as Jordan and Russell put it, "defines the underlying organization of the cognitive processes comprising intelligence, and forms the computational substrate upon which domain-specific capabilities are built. For example, an architecture may provide a generic capability for learning the "physical laws" of the environment, for combining inputs from multiple sensors, or for deliberating about actions by envisioning and evaluating their effects. " (lxxvi)
Although there is at present no agreed-upon theory that defines the range of possible architectures for intelligent system or identifies the best architecture for a given task domain, developing a narratively intelligent system will require selecting the architecture that optimally handles a gamut of interrelated tasks. Depending on whether the system in question involves SPEECH SYNTHESIS or else relies on a graphical interface to situate the user in a virtual environment in which the world of the narrative unfolds, these tasks may include NATURAL LANGUAGE GENERATION and PROCESSING; the use of CAUSAL REASONING to construct or identify chains of causes and effects that form the logical skeleton of the narrative and that participants in the narrated world may form their own (correct or incorrect) mental representations of; related PLANNING activities by virtue of which the system can present or interpret a string of sentences or a sequence of images as the means to convey a story, and also handle participants' plan-based words and actions within the story; and problem solving capabilities that may need to be recruited for stories involving enigmas, strategic omissions, or the like.
In short, the demands placed on a system with genuine narrative intelligence will be considerable (Herman and Young 2000; Mateas and Senger 1999). The effort to meet those demands may help researchers adjudicate between the relative strengths and merits of competing architectures-or at least assess their efficacity for a task environment in which so many facets of intelligence are concurrently set into play. In particular, theorists have distinguished between symbolic vs. connectionist and deliberative vs. reactive architectures. (These are cross-cutting, not symmetrical, distinctions.) As Jordan and Russell note, a strong version of the symbolic approach holds that "algorithmic manipulation of symbol structures is necessary and sufficient for general intelligence" (lxxvi). By contrast, and to jump ahead to the entry on CONNECTIONISM, PHILOSOPHICAL ISSUES, authored by William Ramsey, connectionist theories model intelligence in terms of networks in which there is a "spreading activation" of numerous simple units. As Ramsey writes, "[t]he processing is highly distributed throughout the entire system, and there are no task-specific modules, discrete symbols or explicit rules that govern the operations"(186). Meanwhile, whereas deliberative architectures assume that intelligent behavior is a function of deliberating over possible courses of actions based on the agent's world knowledge and expectations about the results of its actions, reactive systems, such as those associated with BEHAVIOR-BASED ROBOTICS, aim "to implement direct mappings from perception to action that avoid the expensive intermediate steps of representation and reasoning" (lxxviii). Given that producers and interpreters of (and participants portrayed within) narratives seem to rely crucially on just these sorts of "expensive intermediate steps," working to situate the unfolding chain of states, actions, and events against a backdrop of what could have happened or might yet happen in the world of the narrative (Herman forthcoming: chapter 2); given, too, that connectionist models seem to have trouble modelling some of the higher functions of intelligence, including agents' ability to maintain and update a temporal knowledge base consisting of ways the world can be inferred to be changing over time (lxxviii); in the case of systems designed for narratively structured task domains, there seems to be intuitive evidence for the advantages of a deliberative architecture that perhaps exploits connectionist design features, but also retains properties associated with symbolic cognitive modelling.
The longest of the introductory essays is Dan Sperber's and Lawrence Hirschfeld's "Culture, Cognition, and Evolution." The essay's length is explained by its scope. The authors focus on how popluation-level phenomena influence individual cognitive processes both diachronically and synchronically, with superindividual phenomena both shaping the ontogenetic and phylogenetic development of cognitive capacities over time and providing them with specific inputs at any given time. The essay subdivides research in this area into three main categories: cognition in a comparative and evolutionary perspective; culture in an evolutionary and cognitive perspective; and cognition in an ecological, social, and cultural perspective.
The first category includes research on such topics as COGNITIVE ETHOLOGY (i.e., the study of cognitive processes in animals), animal communication, PRIMATE COGNITION, and social play behavior, as well as work on the evolution of human cognitive capacities, as explored in SOCIOBIOLOGY, EVOLUTIONARY PSYCHOLOGY, and other fields. The second category marks a reversal of the perspective adopted in the first; here the research examines not the evolution of cognition in its superindividual setting, but the way culture both manifests (human) cognitive abilities and provides tools for and targets of cognition. Relevant research is being conducted not only in COGNITIVE ARCHAEOLOGY but also by scholars studying HUMAN UNIVERSALS (across what are recognized as highly variable cultural contexts) in fields like linguistics (e.g., Brown and Levinson 1987) and COGNITIVE ANTHROPOLOGY. Topics under investigation include, for example, the universal grounding of cultural models in networks of metaphors; the shared properties of ostensibly distinct systems of COLOR CATEGORIZATION; the cognitive dimensions of WRITING SYSTEMS (and LITERACY) viewed as cultural innovations; and the status of folk classifications of plants and animals (studied under the rubrics of ethnobotany and FOLK BIOLOGY), which can be viewed as providing evidence for transcultural cognitive structures and attributes.
In the third category of research discussed by Sperber and Hirschfeld, theorists examine how ordinary cognition takes place in a socially complex, dynamically evolving, and information-saturated environment. Here again researchers commonly adopt a broadly Vygotskyean approach, emphasizing the sociointeractional roots and requirements of human intelligence. The authors write: "In social species, conspecifics occupy a salient place in [the surrounding] environment, and much of the individual-environment interaction is, in fact, interaction with other individuals. In the human case, moreover, the environment is densely furnished with cultural objects and events most of which have, at least in part, the function of producing cognitive effects. "(cxxiii) Indeed, as Edwin Hutchins's (1995) research (as well as his MITECS entry) on COGNITIVE ARTIFACTS suggests, it is legitimate to question the extent to which cognition, in a social and cultural environment, remains an individual process at all. Cognition can be conceived, instead, as a set of interlocking, socially distributed processes. As Hutchins demonstrates, the cognitive capacities needed to fly a plane are not lodged in the pilot's head; rather, they are distributed among the pilot, the members of the crew, the control panels, and the manuals (or simulation programs) used for flight training purposes. More generally, work in this third category of research insists on the maintenance of standards of ECOLOGICAL VALIDITY in cognitive-scientific inquiry. From this perspective, human intelligence should be viewed as contextually situated and adaptive to an ever-changing environment. The watchword then becomes, not "stimuli" to which perceivers merely react, but rather what the psychologist J. J. GIBSON called AFFORDANCES, i.e., possibilities for action opened up by the overall layout and dynamic unfolding of a context of which humans themselves form a part, and in which they are always actively trying to make their way (cf. Herman and Young in preparation).
Vis-à-vis all three categories of research on the links between culture, cognition, and evolution, narratologists can benefit from greater familiarity with work on how superindividual, population-level phenomena both affect and are affected by human cognition. More than this, though, narrative theorists are in a position to help reframe some of the central questions being explored on this front. In connection with the study of cognition in a comparative and evolutionary perspective, narrative theorists would do well to examine fictional storytelling in a broader context of social play behavior. Producers and interpreters of fictional narratives, like humans and other animals engaged in play, arguably rely on "metamessages" (Bateson 1956) or "contextualization cues" (Gumperz 1982) to distinguish fictitious tales from veridical reports. These cues include framing devices, including the fossilized fiction-marker Once upon a time; recurrent action templates of the sort identified by Vladimir Propp (1968); and any number of word-, proposition-, or episode-level prompts triggering interlocutors to relocate from the here and now of a narrative's telling to the alternative possible world that the storyteller aims fictionally to evoke (Duchan et al. 1995; Ryan 1991). Conversely, though, just as research on social play can help integrate (fictional) storytelling into a broader context of practices of make-believe (cf. Walton 1990), narrative analysts can help identify a fuller range of play behaviors, and a richer repertoire of play-indexing cues, than those hitherto recognized. In particular, narratologists can work toward a finer-grained account of the sociointeractional situations (including those bound up with the insitutions of literature and literacy) that elicit fictional storytelling, as well as a more nuanced understanding of the impact of fiction-making on the situations in which it occurs.
New opportunities for research can also be projected if insights from narrative theory are allowed to cross-pollinate with work on culture in an evolutionary and cognitive perspective and cognition in an ecological, social, and cultural perspective. Although narratologists (e.g., Fludernik 1996; White 1996) have begun to explore diachronic dimensions of written, literary narrative, studying the way it emerged from neighboring generic forms such as the chronicle, the saint's life, and drama, researchers also need to connect the emergence and evolution of oral as well as written narrative to changes in human intelligence itself (cf. MacNeil 1996). Here, indeed, a new field of endeavor comes into view-i.e., a cognitive archaeology of protonarrative, narrative, and quasi-narrative artifacts produced by the world's cultures and subcultures. Further, as CLAUDE LÉVI-STRAUSS (1986) recognized, narrative (e.g., myths) can be characterized as in effect one of the earliest and most powerful cognitive artifacts. Lévi-Strauss argued, first, that myths can be analyzed into paradigmatic classes of "mythemes" (= the smallest meaningful unit of mythical discourse)-although the anthropologist's criteria for identifying mythemes remained notoriously vague and impressionistic. He then suggested that by establishing deep-structural analogies and contrasts between these classes of units, myths allow the cultures in which they circulate to work out cosmological problems (e.g., those relating to the origins of human existence) not solvable by empirical observation alone.
More generally, narratives conveyed by oral as well as written traditions can be viewed as a kind of cognition-enhancing logic in their own right, whereby states and events can be arranged into understandable and manipulable patterns; spatiotemporal relations can be established between regions of experience and between objects contained in those regions; relatively distant or intimate perspectives can be adopted; participants can be assigned roles and situated within networks of beliefs, desires, and intentions; and so on (see Herman forthcoming, Herman and Childs in preparation, and the final section of this essay). In this sense, Hutchins's definition of cognitive artifacts as "physical objects made by humans for the purpose of aiding, enhancing, or improving cognition" (MITECS 126) can be adapted to encompass stories, which are not physical objects as such, but which are intersubjectively verifiable, are "always embedded in larger sociocultural systems that organize the practices in which they are used," and function to "amplify human abilities" (127), notably MEMORY, SPATIAL PERCEPTION, TEMPORAL REASONING, and problem solving. As Hutchins notes, Donald A. Norman (1993) relaxes the definition of cognitive artifacts to include mental as well as material elements. Similarly, I go on to argue in the final section of this essay that, viewed as cognitive artifacts, narratives have a peculiarly double status, existing somewhere in between the material and mental realms, or more precisely in the interplay between semiotic (=material) cues and the global and local mental representations triggered and organized by those cues.
I shift now from what has been a relatively wide-focus discussion of some representative introductory essays in MITECS to an account of specific entries contained in the volume. Again, I examine these sample entries from a dual perspective, exploring both what the cognitive sciences can do for narratology and what narrative theorists can contribute to the study of human cognition. Entries discussed include some of those to which I have already alluded, as well as others not yet mentioned.
Some Representative MITECS Entries
Note that a few of the entries do discuss narrative, suggesting that stories constitute an important target for cognitive-scientific research. (Neither narrative nor story, however, appears in the Subject Index of the volume.) For example, Colwyn Trevarthen argues that narrative plays a crucial role in INTERSUBJECTIVITY, which can be defined as the process by which "mental activity-including conscious awareness, motives and intentions, cognitions, and emotions-is transferred between minds" (415). More specifically, Trevarthen writes that: "Acquired beliefs and concepts of a young child are redescriptions of narrativelike patterns of intention and consciousness that can be shared, without verbal or rational analysis, in a familiar "common sense" world. Narrative expression by rhythmic posturing and gesturing with prosodic or melodic vocalisation, or "mimesis," may have been a step in the prelinguistic evolution of hominid communication and cognition . . . . Pretense and socially demonstrated METACOGNITION is natural in infant play, and imitation of pretend actions and attitudes is essential in the development of imaginative representational play in toddlers of modern Homo sapiens." (416) Trevarthen also points to research suggesting that the acquisition of syntax and word meaning may be fundamentally dependent on narrative-syntax on expressive sequences construed as emotional narratives in game-like rituals (Bruner 1983), and word meaning on "narrative exchanges modulated by dynamic affects and expressions of interest, intention, and feeling deployed by the child and companions in a familiar, common worlds" (416; cf. Locke 1993).
Further, Geoffrey White, in his entry on ETHNOPSYCHOLOGY-i.e., "cultural or 'folk' models of subjectivity, particularly as applied to the interpretation of social action" (286)-discusses the relevance of narrative for "prototypic event schemas," on the basis of which members of social groups can impute emotions and other mental attributes to their cohorts. Commenting on the salience of stories as an organizing principle in ethnopsychological contexts, White notes that narrative genres such as the "life story" provide a crucial resource for representing and communicating social experience-even though there appears to be some evidence that Euro-American cultures "'package' experience in the form of individualized life stories more than many non-Western cultures that do not value or elaborate individual self-narrative" (286). For its part, Patrizia Violi's entry on SEMIOTICS AND COGNITION highlights interconnections between cognitive-scientific research on story grammars (e.g., Mandler 1984; Rummelhart 1980) and structuralist semioticians' work on narrative structure (e.g., Barthes 1977; Greimas 1983; Propp 1968). As Violi puts it, research in both of these areas "aims to individuate an underlying structure in stories and to define the nature of concepts such as state and event, despite the fact that these two traditions seem, unfortunately, to ignore each other" (744).
Notwithstanding a few other indications in the volume that cognitive scientists and narrative theorists have attained a modicum of mutual awareness-for instance, Thomas Ede Zimmerman's discussion of the relevance of narrative-theoretical work on free indirect speech (Banfield 1982; Fludernik 1993) vis-à-vis the study of CONTEXT AND POINT OF VIEW (199)-the great preponderance of textual evidence supports Violi's assessment. By and large, rather than recording past inter-adaptations between two linked areas of research, MITECS entries reveal opportunities for future exchange between narratologists and cognitive scientists. Take, for example, Eleanor J. Gibson's, Karen Adolph's, and Marion Eppler's entry on affordances. As the authors note, the term affordance has an essentially relational meaning; it was coined by J. J. Gibson to capture the reciprocal interactions between an animal and its environment. More precisely, affordances are resources or supports offered by the environment that both embeds and is affected by an animal, which must in turn have the capabilities to perceive and use what the environment affords. Surfaces afford support; objects afford manipulable entities; blizzards afford being frozen; fires afford being warmed; and other animals afford various possibilities for and modes of social interaction (4-5). Further, "[u]tilization of an affordance implies a . . . reciprocal relationship between perception and action. Perception provides the information for action, and action generates consequences that inform perception" (5). From a diachronic perspective, humans and other animals can be viewed as genetically equipped to "instigate exploratory activity that culminates in learning what [their] environment affords for [them]" (6).
As the preceding sketch suggests, ecological psychologists' research on affordances can itself be brought into a productive reciprocal relation with theories about narrative. Indeed, in the case of humans, narrative can be thought of as a sort of "meta-affordance," a global framework for dovetailing perception and action (including communicative action) in an emergent, information-rich environment (Herman and Young in preparation). For instance, in a broadly narrative framework, one's fellow actors can be assigned participant roles that make of (some of) them resources for cooperative social activity. Depending on the narrative stance being adopted, participants can be assigned those roles retrospectively, prospectively, or contemporaneously vis-à-vis the time of narration (Margolin 1999). In addition, narrative enables spatial and temporal information to be woven together into spacetime coordinates defining successively encountered places and events in narrated worlds. In the framework of storytelling, disparate perceptual inputs can be integrated into a unified representation, which the mind "chunks" into a sequence of there-and-thens. Within that sequence, any given there-and-then affords resources for cognition and action because of the way it emerges from some previous situation as well as the way it affects a subsequent one.
Also, the use of stories to structure and thereby comprehend perception-and-action-in-an-environment-which likewise involves tailoring stories to a larger context of perceiving and acting-is an important topic for research on planning and causal reasoning. It may very well be that in running away from you when you come at me with your fist raised, I do so because I have managed to insert your overt actions (taking a particular path of motion, raising your fist, etc.) into a narrative, by which I impute to you motivations and goals and also project the likely outcome of your actions should I take no countervailing steps. In sum, a question for future cognitive-scientific inquiry (e.g., COMPARATIVE PSYCHOLOGY and cognitive ethology) is whether narratives afford humans with affordances-opportunities for perceiving and acting-of a different sort than those available to other animal species.
Tools from narrative analysis can help reframe issues discussed in other MITECS entries as well, including those devoted to CASE-BASED REASONING AND ANALOGY, HUMAN-COMPUTER INTERACTION, and SITUATEDNESS/EMBEDDEDNESS-to mention just a few relevant entries. Ronald Loui defines case-based reasoning (CBR) as "a style of designing a [computer] system so that thought and action in a given situation are guided by a single distinctive prior case (precedent, prototype, exemplar, or episode)" (99). CBR works with a set or corpus of past cases, i.e., a case base, and uses a two-stage reasoning process to determine a source case relevant to a target case. The first stage involves retrieving, or finding the appropriate source case, prototype, or precedent; the second stage involves revising/reusing, or working out the appropriate inferences to be drawn about the target case-on the basis of how the source is mapped onto the target. As Loui remarks, CBR can be contrasted with induction. Specifically, "CBR differs from induction because induction derives its power from the aggregation of cases, from the attempt to represent what tends to make one case like or unlike another. [By contrast,] CBR derives its power from the attempt to represent what suffices to make one case like or unlike another" (100).
In this connection, Turner's (1996) concept of "parabolic projection"-the parable-like mapping of familiar stories into novel situations that they thereby make comprehensible and manageable-itself consitutes an important analogue for CBR, and may throw light on the operations of retrieving and revising/reusing that are CBR's basic components. At the very least, what triggers the projection of one narrative and not another into a particular set of circumstances, and exactly how that projection helps make sense of the target circumstances, are questions mirroring those being addressed in research on CBR. A stronger claim can be made here, though. Theories of narrative may in fact help illuminate the foundational concept of CBR-namely, the idea of "case." Arguably, narrative is a basic resource for structuring the flow of experience into prototypic situations or cases and noteworthy, reportable departures from those prototypes; i.e., stories provide cognitive and communicative templates that can be used to establish mapping relations beween emergent data and situations already undergone, cases already encountered. A story's narrativity (= the degree to which it is amenable to being processed as a narrative) increases in direct proportion with the unusualness of the situation it reports vis-à-vis prior, prototypic situations (cf. Herman 1997, forthcoming: chapter 3). Hence studies of narrative processing may be relevant for what would seem to be the crucial test for CBR-i.e., situations in which the real or perceived discrepancy between source and target cases is such that it makes the operations of retrieving and revising/reusing particularly difficult to effectuate.
Meanwhile, in his entry on human-computer interaction (HCI), James D. Hollan defines the field as the study of "how people design, implement, and use computer interfaces" (379). Hollan further notes that, insofar as they are designed to aid cognition and simplify tasks, interfaces can be classified as cognitive artifacts, with today's highly interactive interfaces mediating a "redistribution of cognitive tasks between people and machines" (379). It is worth quoting Hollan at greater length: "Two features distinguish interfaces from other cognitive artifacts: they provide the most plastic representational medium we have ever known, and they enable novel forms of communication. Interfaces are plastic in the sense that they can readily mimic representational characteristics of other media. This plasticity in combination with the dynamic character of computation makes possible new interactive representations and forms of communication that are impossible in other media." (379) In fact, the plasticity of interactive human-computer interfaces affords opportunities for research that have only begun to be explored (see, e.g., Herman and Young 2000 and in preparation; Young and Herman 2000; and the website for Project Mimesis, which can be accessed at http://mimesis.csc.ncsu.edu). Whereas current interfaces to information technology can sometimes be quite difficult to use, particularly for young students and others who do not have much experience with software applications, stories might enable the design and implementation of more familiar and natural-more immersive-interfaces for people who are trying to accomplish learning, research-related, or other tasks in computer-mediated environments.
In educational settings in particular, a major theoretical as well as engineering challenge is to develop storylike environments that simulate for student users the experience of being caught up in a gripping narrative. Thanks to significant advances in the graphics rendering capabilities of off-the-shelf computer and video games, virtual microworlds-for example, the meadhall in Beowulf-can be equipped with authentic-looking settings, artifacts, and characters. Narratological research on character and on modes of perspective-taking can enrich system design, enabling users to select various sorts of participant roles and to adopt more or less externalized or internalized views on the actions and events that their own input helps trigger. By the same token, AI research on planning can be used to structure and monitor users' interactions with the system over time. The system will allow user inputs that are preconditions for the unfolding of the story-e.g., Beowulf's heroic defense of the meadhall against Grendel's predatory attacks requires that the warrior first arrive at Heorot-but it will block choices that would introduce fatal levels of incoherence into the emerging narrative. A user playing the role of Grendel, for example, could not suddenly opt to become a nonviolent, altruistic supporter of the Danes, on pain of eliminating the conflict that is basic to the narrative and that makes the story tellable in the first place. It is not just that the "telling" of such a narrative would be distributed between human and machine. What is more, the mode of distribution of intelligence will have properties essentially narrative in nature, with the interface having been designed in such a way as to mimic the unfolding of a plot, and to immerse the user in the microworld by inserting him or her into in a particular character's vantage-point on that environment.
Further, research in narrative theory has a direct bearing on issues discussed in Brian Cantwell Smith's entry on situatedness/embeddedness in the context of computational intelligence. As Smith points out, "[t]he situated movement . . . views intelligent human behavior as engaged, socially and materially embodied activity, arising within the specific concrete details of particular (natural) settings, rather than as an abstract, detached, general-purpose process of logical or formal ratiocination" (769; cf. Agre 1997). In other words, in contrast to GOFAI (or "good old fashioned artificial intelligence"), which views cognition as individual, rational, abstract, detached, and general, the situated approach argues that cognition is fundamentally social, embodied (= bound up with material aspects of agents' bodies), concrete (= subject to physical constraints of realization and circumstance), located (= both dependent on and enabled by particular contexts), engaged (= marked by an evolving interplay between agents and their environment), and specific (= oriented around contingent facts grounded in highly variable circumstances).
Viewed thus, cognition shares many of the properties that Arthur C. Danto (1985) associates with what he terms narrative explanations, which Danto contrasts with so-called covering-law explanations described by philosophers of science. A narrative explanation does not take the form of instance-of-a-physical-law (e.g., water freezes at thirty-two degrees Fahrenheit and zero degrees Celsius; X is water; therefore X will freeze at . . .); rather, story-based explanations focus on a more or less extended chain of paticular circumstances bound together by uniquely instantiated causal and chronological principles (e.g., a pond froze in Indiana in December, 1995, after a spate of six very cold days). Such explanations can involve atomic or molecular narratives, depending on how many "links" of the causal-chronological chain fall within their scope. The story of the pond's freezing involves an atomic narrative; the story of the decline and fall of the Roman Empire, a molecular narrative, or rather an extended sequence of molecular narratives (cf. Herman forthcoming: chapter 2). This shift from a metric of generality to a metric of size aligns narrative explanation with the situated approach to cognition: narratives may include smaller or larger sets of particular circumstances, but they do not abstract away from those particularized contexts to isolate what they all have in common. By the same token, narratives center around the interplay between located, embodied agents and their environment, characteristically focusing on the conflicting plans of social actors trying to bring contingencies of their shared environment into line with their respective beliefs, desires, and intentions. Narrative thinking, in short, can be construed as inherently situated, and hence rich with implications for theories of situated cognition.
Conversely, however, I should reemphasize that MITECS entries discuss issues that can provide new directions for narrative-theoretical research. Relevant entries include, among many others, those devoted to the binding problem, DISCOURSE, and JUDGMENT HEURISTICS. Indeed, the entry on judgment heuristics, together with those on DOMAIN SPECIFICITY and the MODULARITY OF MIND, can help reshape current thinking about what narrative is and how it should be studied.
John Hummel's entry on the binding problem defines it as, most basically, the problem of representing conjunctions of properties. For example, to be able to detect a solid, vertical red line among solid and dotted vertical blue lines and solid and dotted diagonal red lines, perceivers must be able to bind each line's color to its orientation as well as its solid or dotted quality. Further, to parse the sentence Walter pushed Dan, language users must be able to bind Walter to the thematic role of Agent and Dan to the thematic role of Patient. More generally, as Hummel notes, "[b]inding lies at the heart of the capacity for symbolic representation" (85), with narrative modes of representation being no exception. Narrative understanding requires, in particular, binding actions and occurrences to temporal segments of the narrated world that can precede, follow upon, or be indeterminately located in time relative to one another. Comprehending a story also requires binding actions and events to particular spatial regions of the world the narrative evokes (cf. Emmott 1997), as well as binding participants to types of processes (material, mental, verbal, etc.) that they may initiate or undergo, thereby taking on roles and relations that the early narratologists tried to describe under the heading of actants (Greimas 1983; cf. Herman forthcoming: chapter 4). In re-exploring these and other aspects of narrative processing, narratologists should consider the work on static vs. dynamic binding-and on the links between binding and ATTENTION and working memory-that Hummel synopsizes in his entry.
Equally consequential for narrative theory is Craige Roberts's entry on linguistic, sociolinguistic, and cognitive-linguistic approaches to the study of multisentential units of language, i.e., discourse. As Roberts notes, discourse can be characterized in several ways: as "a type of event, in which human agents engage in a verbal exchange"; as the linguistic content of that exchange, a sequence of words with their associated syntactic, semantic, and prosodic structures; and also as "the more complex structure of information that is presupposed and/or conveyed by the interlocutors during the course of the discourse event in view of the explicit linguistic content of the exchange" (231). A substantial body of scholarship has grown up around each of these dimensions or profiles of discourse, and narratologists-the investigators, after all, of narratively organized discourse in particular-can gain much from a fuller acquaintance with this domain of research (cf. Herman 1999).
Sociolinguistic research in the traditions of conversation analysis and interactional sociolinguistics can throw light on the exchange-structure of narrative discourse, its status as an interactional achievement of storytellers/narrators and interlocutors/readers. Specifically, the achievement of narrative is made possible by verbal cues anchored in turn-taking protocols-protocols required for the initiation and maintenance of the more or less extended turns at talk required to tell a story. Meanwhile, both the linguistic and the informational profiles of narrative discourse can be illuminated with the help of research on the discourse functions of syntax-i.e., the way topicalizations (This book I dislike), clefted constructions (It is this book that I dislike), and other syntactic options affect discourse processing (cf. Birner and Ward 1998). Also relevant is research on PROSODY AND INTONATION as cues to meaning in discourse (cf. Couper-Kuhlen 1986; Gumperz 1982: 100-29); on processes of centering that, bound up with particular classes of discourse cues, bring certain topics into and out of FOCUS over time (Grosz and Sidner 1986; Walker, Joshi, and Prince 1998); and on the ways in which language-users (including narrators) are able to communicate more than they explicitly say, thanks to general and basic principles of rationality discussed by H. P. Grice in connection with what he called conversational IMPLICATURE (1989; cf. Pratt 1977).
What is more, Baruch Fischhoff's entry on judgment heuristics affords new insight into a problem recognized early on by Roland Barthes in his 1966 "Introduction to the Structural Analysis of Narratives" (Barthes 1977). Barthes observed that "the mainspring of narrative is precisely the confusion of consecution and consequence . . . in which case narrative would be a systematic application of the logical fallacy denounced by Scholasticism in the formula post hoc, ergo propter hoc [= after this; therefore, because of this]" (1977: 94). In other words, narrative understanding depends fundamentally on pragmatic principles according to which interpreters assume that if Y is mentioned after X in a story, then X not only precedes but also causes Y. Analogous principles account for language-users' ability to "read in" temporal relations in the case of conjunctions that do not contain explicit time-indices. Thus, most speakers of English would agree that the string Getting married and having a baby is the thing to do differs from Having a baby and getting married is the thing to do, even though there is nothing "in" the connective and to mark a particular temporal ordering of the two conjuncts contained in both strings. In the case of narrative in particular, such pragmatically based principles of comprehension are indispensable to tellers as well as listeners/readers-to tellers because they can safely leave things unstated that it would otherwise take far too much time and effort to spell out, and to listeners/readers because they can make sense of such abbreviated reports, which would otherwise remain hopelessly elliptical and opaque. If in telling a story I recount that someone got into bed and then report that the person in question fell asleep, I need not elaborate the causal as well as chronological link between these two events but can instead assume that you will use your general world- knowledge about beds as places for sleeping to sketch in the link yourself. (See the entries on the FRAME PROBLEM and SITUATION CALCULUS for information about attempts to replicate such powerful inference-drawing capabilities in contexts of computational intelligence.)
Fischhoff's entry on judgment heuristics suggests that these pragmatic dimensions of narrative understanding may be aligned with more general "coping strategies"-strategies by virtue of which people arrive at determinations based not on exact calculations but on heuristics, or rules of thumb. Heuristics of this sort, influentially explored by Amos Tversky and Daniel Kahneman (1971, 1974; cf. also Kahneman, Slovic, and Tversky 1982), enable people to determine the probability of rain, the size of a crowd on the street, or the time needed to write a longish review-article-though they also lead to systematic errors that researchers have technically defined as the biases attendant upon distinct kinds of judgment heuristics. For example, and as Fischhoff notes, Tversky and Kahneman (1971) found evidence to support the claim that "people expect future observations of uncertain processes to be much like past ones, even when they have few past observations to rely on. Such people might be said to apply a 'law of small numbers,' which captures some properties of the statistical 'law of large numbers' but is insufficiently sensitive to sample size" and yields predictable and sometimes very unfortunate biases (often informally referred to as STEREOTYPING) (423). In later research, Kahneman and Tversky subsumed this tendency under a broader representativeness heuristic (or rather meta-heuristic), and went on to identify two other meta-heuristics-i.e., availability and anchoring and adjustment (424). The availability meta-heuristic involves judging an event to be likely based on how easy it is to remember examples of such an event or to imagine the event happening. Meanwhile the anchoring and adjustment meta-heuristic predisposes people to adjust too little once they have been given a value to which they are initially "anchored" but which they are then required to revise. For example, if you are told that it is fifty degrees Fahrenheit outside when in reality it is thirty degrees, you are likely to maintain that it is quite a bit warmer than thirty degrees when you try to adjust for actual conditions.
The question that narrative theorists can pose, by integrating Barthes's insights and the work on judgment heuristics reviewed by Fischhoff, is whether narrative itself constitutes a meta-heuristic. It may in fact be the case that human beings use stories as a kind of assemblage of inter-related rules of thumb, each with attendant biases that need to be identified and studied. People do seem to rely on a story-based rule of thumb when they bind strings of successively occurring events into causal and chronological wholes-e.g., I am out in a storm with a group of my peers and only I am struck by lightning; therefore I have earned the ire of the gods or, in another story, I in particular have been brought low by the turning of Fortuna's wheel. Likewise, in western cultures at least, a basic feature of our (egocentric) existence involves casting ourselves as protagonists in an unfolding life story (Linde 1993), in which our loved ones, friends, and acquaintances are ascribed more or less cooperative or antagonistic roles as subsidiary participants. At a more global level, and to revert once again to Turner's (1996) idea of parabolic projection, source stories do indeed seem to provide a heuristic for interpreting unfamiliar circumstances onto which they are projected as targets or explananda-whether or not the fit between source and target is as precise as people may then go on to assume. At the very least, there are grounds for future inquiry into the heuristic or meta-heuristic functions of stories, as well as the biases that might systematically follow from people's predisposition to find the seeds of story in unrelated (or non-narratively related) circumstances and events.
Similarly, Susan A. Gelman's entry on domain specificity, together with Arnette Karmiloff-Smith's entry on the modularity of mind, raises questions that are vitally important for narrative-theoretical research. Indeed, like Fischhoff's account of judgment heuristics, Gelman's and Karmiloff-Smith's entries introduce issues that have a bearing on the very foundations of future research on stories. Building on work (e.g., Fodor 1983; Hirschfeld and Gelman 1994) that has emerged as central to recent cognitive-scientific inquiry, Gelman notes that "[c]ognitive abilities [can be described as] domain-specific to the extent that the mode of reasoning, structure of knowledge, and mechanisms for acquiring knowledge differ in important ways across distinct content areas" (238). Candidate domains for content-bound modes of cognition include language, number processing, face perception, and spatial reasoning. Further, Gelman points out that theories about the MODULARITY OF MIND assume domain-specificity in the previous sense.
As Karmiloff-Smith discusses in her entry, which in fact goes on to challenge some the tenets of modularity theory, a module can be defined as a fast, automatic, and mandatory special-purpose processing device that takes input from sensory transducers, operates on it in an informationally encapsulated way (= a way that is not influenced by and does not influence other parts of the mind while the input is being worked on), and then "outputs data in a common format suitable for central, domain-general processing" (MITECS: 558). Initially, as Gelman remarks (238), linguists like Noam Chomsky (1988; cf. Pinker 1994) pointed to evidence for the neurological localization of linguistic abilities, as well as humans' rapid acquisition of language despite limited environmental data (cf. the entry for POVERTY OF THE STIMULUS ARGUMENTS), to argue for the existence of independent modules of linguistic knowledge (e.g., syntax and phonology). Subsequently, however, Fodor generalized the approach, extending it from linguistic to more broadly cognitive abilities. In broadening the modularity approach, Fodor drew a distinction between central logical processes and modularized perceptual systems. As Gelman writes: "Like the visual system, by Fodor's analysis, modules are innately specified, [with] their representational outputs [remaining] insensitive to revision via experience. Experience provides specific inputs to modules, which yield mandatory representations of inputs. Certain experiential inputs may be necessary to trigger the working of each module in the first place, but the processes by which the module arrives at its representations are mandatory rather than revisable." (238-39)
It seems appropriate to conclude this sampling of MITECS entries with a set of questions that recent work on domain-specificity and the modularity of mind makes it possible-indeed, imperative-for narrative theorists to pose in future research. Is narrative knowledge (i.e., the knowledge needed to tell and understand stories) itself domain-specific? I.e., if stories do in fact constitute a meta-heuristic, are narrative-based rules of thumb grounded in some innately specified, automatic, and informationally encapsulated module or cognitive domain, in which processing strategies, distinct from those geared to other content areas, are autonomously at work? To put the same question still otherwise, can the apparently deeply entrenched status of story-based judgment heuristics be explained by the domain specificity of narrative knowledge? Is it just because the representational outputs of the "narrative module" remain insensitive to revision via experience that they generate narrative templates to whose use in everyday thinking people feel irresistibly predisposed-even when they result of systematic biases like the post hoc, ergo propter hoc effect that Barthes discussed in his "Introduction"? Further, if narrative is indeed a special-processing device of this kind, how exactly is its output used by the central, domain-general logical processes hypothesized by Fodor? How is the output of the narrative module integrated with that of the other modules running in parallel with it, e.g., language, vision, or spatial reasoning? And in what way, precisely, do Fodor's "sensory transducers" filter experiential input such that they can trigger the working of the narrative module in the first place?
Narratology and the Architecture of Inquiry
Up to now I have used MITECS as a platform for reassessing the relations between narratology and the cognitive sciences. More precisely, in pointing to mutually illuminating developments in these two areas of research, I have been suggesting that narratology should be classed among the cognitive sciences. In this concluding section of my essay I should like to articulate and argue for that claim in a different way, shifting my focus from MITECS itself to the more general issues raised by its cross-disciplinary profile. Here I provide a synopsis of arguments developed in more detail in my forthcoming study of "story logic" (Herman forthcoming). In particular, revisiting the way linguistic models have been used by narratologists since the beginnings of structuralist narrative theory, and comparing this cross-fertilization with narrative analysts' more recent borrowing of concepts and methods from cognitive science, I suggest the advantages of an alternative approach. My claim, here as in Story Logic, is that both narrative theory and language theory should instead be viewed as resources for-elements of-the broader endeavor of cognitive science. The result: a jointly narratological and linguistic approach to stories construed as strategies for building mental models of the world.
Narrative Theory and Linguistic Science
It would be hard to dispute that linguistic models have had a major impact on narrative theory over the past three or four decades-that is, from its very inception. In founding the discipline of narratology (or at least naming it), Todorov's 1969 study of Boccaccio's Decameron borrowed categories from traditional grammars to compare narrated entities and agents with nouns, actions and events with verbs, and properties with adjectives (Todorov 1969). Gérard Genette (1980) drew on the same grammatical paradigm in using tense, mood, and voice to characterize the relations between the narrated world, the narrative in terms of which it is presented, and the narrating that enables the presentation. Before Genette and Todorov, Lévi-Strauss (1986) had patterned his concept of mythemes, however quixotically, on Troubetzkoy's, Saussure's, and Jakobson's understanding of the phoneme as a bundle of distinctive features. And whereas Bremond (1973, 1980) thought of himself as working to build a logic rather than a grammar of narrative, Barthes's 1966 "Introduction" started from the premise that discourse is the object of a "second linguistics" (Barthes 1977: 83), a linguistics for units of language beyond the sentence, in the context of which "[t]he general language [langue] of narrative is one (and clearly only one) of the idioms apt for consideration" (84).
The broad influence of linguistic models on narrative theory, then, is undeniable. But the precise nature, extent, and consequences of this influence-some might say contagion-remain open to question. Indeed, almost as soon as the early narratologists followed other structuralists in conferring on linguistics the status of a "pilot-science" (Dosse 1997: I, 59-66), metatheoretical inquiry into the the relations between linguistic and narratological models became a basic research activity, a gesture in part constitutive of the field. There was, it is true, a brief, heady period of what might be called methodological utopianism, a fervent if short-lived belief in the power of linguistic models to revolutionize the study of narrative and more broadly literary and cultural phenomena. Such utopianism can be found in Barthes' 1966 "Introduction," and it is even more palpably evident in his programmatic essay on "The Structuralist Activity" (Barthes 1971b), published two years earlier (cf. Herman under review). Almost immediately, however, the goal of narrative theorists modulated from a more or less uncritical celebration of linguistic paradigms into an effort to adapt certain kinds of models for certain descriptive and explanatory tasks.
Remarking that theorists such as Barthes and Todorov had failed to identify "with precision the basic structural units of a story" (11), Gerald Prince's 1973 Grammar of Stories argued that researchers could build a more explicit and more complete model of narrative by replacing traditional grammatical categories with transformational-generative paradigms. Similarly, in 1975, Jonathan Culler's Structuralist Poetics comprehensively reexamined the "The Linguistic Foundation" (3-31) of work by theorists such as Barthes, Genette, Greimas, and Jakobson. Culler also devoted a chapter to the role of "Linguistic Metaphors in Criticism" (97-109, my emphasis), the title of his chapter suggesting not a kneejerk assimilation of linguistic models and methods, but rather a reflexive adaptation of certain elements of linguistic theory for certain kinds of narratological and literary-theoretical problems. 1979 saw Marie-Laure Ryan drawing on developments in generative semantics to sketch a second-order critique of Prince's syntactically-oriented story grammar, with Ryan already suggesting ways to refine strategies for refining what was itself a rethinking of structuralist narratology! During the 1980s and 1990s, this process of narratological autocritique (and auto-autocritique) accelerated, as exemplified by the diverse contributions of Lubomír Dolezel (1998), Monika Fludernik (1996), Manfred Jahn (1997, 1999), Uri Margolin (1986, 1990, 1999), Thomas Pavel (1985, 1986, 1989), Shlomith Rimmon-Kenan (1983, 1989), and Ryan (1991), among others. All of these researchers, despite the diversity of their orientations, arguments, and examples, have addressed core narratological problems by trying to ascertain what sorts of linguistic models can most fruitfully be brought to bear on them.
Dolezel (1998), Pavel (1986), and Ryan (1991), for instance, have sought to overturn the structuralist moratorium on referential issues, using tools from model-theoretic or possible-worlds semantics to characterize the world-creating properties of narrative discourse. Meanwhile, Fludernik (1996) has drawn on methods for analyzing oral narrative to argue for a gradualist approach to the study of stories; for her a continuum stretches between the tales exchanged in face-to-face interaction and the most avant-garde literary narratives, with both conversational participants and readers of postmodern fiction using telling, viewing, acting, and experiencing parameters to organize their understanding of an unfolding narrative-i.e., to process the spoken or written discourse as narrative in nature. The same emphasis on cognitively-based frames and parameters informs Jahn's recent efforts to fashion a cognitive narratology. For Jahn (1997), higher-order knowledge representations or frames enable interpreters of stories to disambiguate pronominal references, decide whether a given sentence serves a descriptive or a thought-reporting function, and, more generally, adopt a top-down as well as a bottom-up approach to narrative processing. Readers attach emergent details about a character, situation, or event to a global interpretive frame (e.g., authorial narration, or figural narration) until such time as the details force a more or less conscious reanalysis of the narrative from the perspective of a different or more expansive frame (Jahn 1999).
Approaches such as Jahn's and Fludernik's thus call for updating and enriching narratological theories by incorporating models and tools from discourse analysis, linguistic pragmatics, and cognitive linguistics. By contrast, in an approach that resituates narratology as a subdomain of cognitive-scientific research, the most pressing task becomes, not characterizing the role of linguistic or cognitive-linguistic models in narrative theory, but rather reorganizing the study of language and narrative in ways that allow for a new interlocking of methodologies, and new synthesis of research methods and aims. This sort of redrawing of the architecture of inquiry is, I contend, no trivial pursuit. For one thing, it suggests that narrative theorists should combine several methods of linguistic analysis to study aspects of narrative understanding. For another, it alters and enlargens the horizons of linguistic research itself, recasting language as a crucial interface between narrative and cognition. My final subsection spells out these last claims in somewhat more detail.
Story Logic as a Target for Future Research
In an approach that includes narratology among the cognitive sciences, a major target of narrative analysis will be the process by which interpreters reconstruct storyworlds, i.e. the worlds (= configurations of participants, objects, and places and sequences of states, events, and actions) evoked by cues contained in narratives. Story logic refers to both production and interpretation issues associated with this process of storyworld (re)construction. Hence the dual profile of the logic in question. It is not just that stories have a logic consisting of design principles that, as discussed below, operate at relatively local as well as relatively global levels of narrative structure. More than this, narrative constitutes a logic in its own right, providing human beings with one of their primary resources for organizing and comprehending experience. Put briefly, story logic enables people both to build and to comprehend storyworlds, by virtue of which experience itself can be structured and rendered cognizable, manipulable, liveable.
In turn, storyworlds can be recharacterized as mental models of who did what to and with whom, when, where, why, and in what fashion in the world to which recipients relocate-or make a deictic shift (Duchan et al. 1995)-as they work to comprehend a narrative. I thus use the term world (and storyworld) in a manner more or less analogous with cognitive linguists' use of the term discourse model. A discourse model can be defined as a global mental representation enabling interlocutors to draw inferences about items and occurrences either explicitly or implicitly included in a discourse (Emmott 1997; Green 1989; Grosz and Sidner 1986; McKoon et al. 1992; Webber 1979). By the same token, and like Jahn's cognitive frames, storyworlds-or models for understanding narrative discourse in particular-function in both a top-down and a bottom-up way during narrative comprehension. They guide readers to assume that jets, cell phones, and plasma guns do not exist in the world of Madame Bovary (Flaubert 1992). But they are also subject to being updated, revised, or even abandoned with the accretion of textual cues, as when the reader of a narrative told by an unreliable narrator gradually realizes that the storyworld is not at all what its teller portrays it as being.
Fundamentally, then, narrative comprehension is a process of (re)constructing storyworlds on the basis of textual cues and the inferences that they make possible. For heuristic purposes, this process can be decomposed into two broad modeling tasks, each with its associated subtasks, and each requiring a synthesis of cognitive-scientific, narratological, and linguistic paradigms for its description and analysis. The first task is that of establishing, at a relatively local level, an inventory of what can be called principles for narrative microdesigns. Such principles bear on interpreters' sense of what is going on-what needs to be mentally modeled-during comparatively short stretches of the unfolding storyworld. These "small" design principles include coding strategies used to apportion particular facets of storyworlds into states, events, and actions; they also encompass the fashioning of action structures in terms of which individual behaviors can be identified as elements of somewhat larger sequences of occurrences (e.g., pursuing a goal, avoiding an antagonist, aiding and abetting an ally, or instigating or resisting a behavioral or conceptual trend). Other microdesign principles bear on the roles and relations of storyworld participants, whose dialogues and styles help index their emergent attitudes, stances, and interrelations.
The second set of modelling tasks encompasses principles for narrative macrodesigns. Relevant here are "large" design principles determining not so much the individual constituents or localized features as the overall contours, the dominant "feel," of the storyworld being mentally modelled. Narrative macrodesigns determine, for example, whether narrated events can be located definitely in time, or whether their temporalization is left strategically inexact, thanks to fuzzy or indeterminate temporal ordering. Further, narrative understanding involves complementary processes of spatialization. Although many theorists of narrative have accentuated its temporal properties-such that a story, for Seymour Chatman (1990), can defined as a sequentially organized representation of a sequence of events-comprehending a narrative also requires spatializing or cognitively mapping the storyworld it conveys. Making sense of a story entails situating participants and other entities in emergent networks of foreground-background relationships. Story comprehension entails, too, mapping the trajectories of individuals and objects as they move or are moved along narratively salient paths.
Intimately related to such processes of spatialization are those of perspective-taking. As one of the principal means of adopting vantage-points on people, places, things, actions, and events, stories index modes of perspective-taking by way of personal pronouns; definite and indefinite articles; verbs of perception, cognition, and emotion; tenses and verbal moods; and evaluative lexical items and marked syntax. Another macrodesign principle can be called contextual anchoring, or the process by which cues in narrative discourse trigger recipients to establish a more or less direct or oblique relationship between the stories they are interpreting and the contexts in which they are interpreting them. Contextual anchoring, enabled by mechanisms of address, deictic references, and other textual prompts, is thus a way of characterizing aspects of the interface between stories and their interpreters.
As should be apparent from this thumbnail sketch, in order to begin conducting even a partial inventory of local and global principles for storyworld design, narrative analysts must address a whole cluster of problems, each quite formidable in its own right. What distinguishes an event from a state? What, exactly, constitutes an action? How do narratives at once depend on and enable interpretation of events as goal-directed actions? On the basis of what cognitive mechanisms do readers or listeners of narratives form inferences about sequential relationships between actions, and in what textual features are those inferences anchored? Does narrative itself (operating in a feedback loop of some sort) help shape people's ability to emplot their experiences, to mold their worlds into storyworlds? How do inferences about participant roles bear on the process of narrative comprehension, and conversely, how do speech representations in narrative bear on inferences about participant roles? What sorts of textual prompts cue interpreters to draw inferences about the spatiotemporal profile of storyworlds, and how are those inferences additionally constrained by modes of perspective-taking also indexed by cues in the discourse? Why does contextual anchoring operate differently in (certain styles of) second-person narration than in first- or third-person narratives featuring a fully characterized intradiegetic narratee?
Addressing these sorts of questions will require incorporating insights afforded by theoretical and empirical work in the cognitive sciences-work to which MITECS provides such a comprehensive and up-to-date guide. But by the same token, and as I hope to have demonstrated in this essay, the theories, data, and insights reported in MITECS will be altered and enriched when they are refocused around research hypotheses taking shape within what is now a new narratology-a narratology that the cognitive sciences have reciprocally begun to make possible. Insofar as narratology both requires emergent topics of cognitive-scientific study to frame some of its most important questions and also suggests new strategies for framing the questions being explored by practicing cognitive scientists, it deserves a place alongside the other subfields so richly documented in MITECS. There are, in short, compelling reasons for reclassifying narratology as a cognitive science.
Agre, Philip E. 1997. Computation and Human Experience. Cambridge: Cambridge University Press.
Banfield, Ann. 1982. Unspeakable Sentences: Narration and Representation in the Language of Fiction. Boston: Routledge and Kegan Paul.
Barthes, Roland. 1971a. "Action Sequences." In Joseph Strelka, ed., Patterns of Literary Style, 5-14. University Park, PA: Pennsylvania State University Press.
-----. 1971b. "The Structuralist Activity." In Hazard Adams, ed., Critical Theory Since Plato, pp. 1196-99. San Diego: Harcourt Brace Jovanovich.
-----. 1974. S/Z. Trans. Richard Miller. New York: Hill and Wang.
-----. 1977. "Introduction to the Structural Analysis of Narratives." In Image Music Text, pp. 79-124. Trans. Stephen Heath. New York: Hill and Wang.
Bartlett, Frederic C. 1932. Remembering. Cambridge: Cambridge University Press.
Bateson, Gregory. 1956. "The Message 'This is Play.'" In B. Schaffner, ed., Group Processes: Transactions of the Second Conference, pp. 145-256. New York: Josiah Macy Foundation.
Birner, Betty J., and Gregory Ward. 1998. Information Status and Noncanonical Word Order in English. Amsterdam: John Benjamins.
Bremond, Claude. 1973. Logique du récit. Paris: Seuil.-----. 1980. "The Logic of Narrative Possibilities." Trans. Elaine D. Cancalon. New Literary History 11: 387-411.
Brown, Penelope, and Stephen C. Levinson. 1987. Politeness: Some Universals in Language Usage. Cambridge: Cambridge University Press.
Bruner, Jerome. 1983. Child's Talk: Learning to Use Language. New York: Norton.
-----. 1986. Actual Minds, Possible Worlds. Cambridge, MA: Harvard University Press.
-----. 1991. "The Narrative Construction of Reality." Critical Inquiry 18: 1-21.
Chatman, Seymour. 1990. Coming to Terms: The Rhetoric of Narrative in Fiction and Film. Ithaca: Cornell University Press.
Chomsky, Noam. 1988. Language and Problems of Knowledge. Cambridge, MA: MIT Press.
Cohn, Dorrit. 1978. Transparent Minds: Narrative Modes for Presenting Consciousness. Princeton: Princeton University Press.
Couper-Kuhlen, Elizabeth. 1986. An Introduction to English Prosody. London: Edward Arnold.
Culler, Jonathan. 1975. Structuralist Poetics: Structuralism, Linguistics, and the Study of Literature. Ithaca: Cornell University Press.
Danto, Arthur C. 1985. Narration and Knowledge (including the integral text of Analytical Philosophy of History). New York: Columbia University Press.
Defoe, Daniel. 1981. Moll Flanders. New York: Signet Classic.
Dolezel, Lubomír. 1998. Heterocosmica: Fiction and Possible Worlds. Baltimore: Johns Hopkins University Press.
Dosse, François. 1997. History of Structuralism. Vols. 1 and 2. Trans. Deborah Glassman. Minneapolis: University of Minnesota Press.
Duchan, Judith F., Gail A. Bruder, and Lynne E. Hewitt, eds. 1995. Deixis in Narrative: A Cognitive Science Perspective. Hillsdale, NJ: Lawrence Erlbaum.
Emmott, Catherine. 1997. Narrative Comprehension: A Discourse Perspective. Oxford: Oxford University Press.
Flaubert, Gustave. 1992. Madame Bovary: Patterns of Provincial Life. Trans. Francis Steegmuller. New York: Vintage Books.
Fludernik, Monika. 1993a. The Fictions of Language and the Languages of Fiction. London: Routledge.
-----. 1996. Towards a "Natural" Narratology. London: Routledge.
Fodor, Jerry A. 1983. Modularity of Mind. Cambridge, MA: MIT Press.
Frawley, William. 1997. Vygotsky and Cognitive Science: Language and the Unification of the Social and Computational Mind. Cambridge, MA: Harvard University Press.
Genette, Gérard. 1980. Narrative Discourse: An Essay in Method. Trans. Jane E. Lewin. Ithaca: Cornell University Press.
Greimas, Algirdas-Julien. 1983. Structural Semantics: An Attempt at a Method. Trans. Danielle McDowell, Ronald Schleifer, and Alan Velie. Lincoln: University of Nebraska Press.
Green, Georgia M. 1989. Pragmatics and Natural Language Understanding. Hillsdale, NJ: Lawrence Erlbaum.
Grice, H. P. 1989. Studies in the Way of Words. Cambridge: Harvard University Press.
Grosz, Barbara and Candace Sidner. 1986. "Attention, Intentions, and the Structure of Discourse." Computational Linguistics 12: 175-204.
Gumperz, John J. 1982. Discourse Strategies. Cambridge: Cambridge University Press.
Herman, David. 1997. "Scripts, Sequences, and Stories: Elements of a Postclassical Narratology." PMLA 112: 1046-1059.
-----. 1999. "Towards a Socionarratology: New Ways of Analyzing Natural-language Narratives." In David Herman, editor, Narratologies: New Perspectives on Narrative Analysis, pp. 218-246. Columbus: Ohio State University Press.
-----. Forthcoming. Story Logic: Problems and Possibilities of Narrative. University of Nebraska Press.
-----. Under review. "Sciences of the Text."
Herman, David, and R. Michael Young. 2000. "Narrative Structure in Intelligent Tutoring Systems." Paper delivered at the Society for the Study of Narrative Literature; Atlanta, GA, April 2000
-----. In preparation. "Narrative Affordances in Intelligent Exploratory Environments."
Herman, David, and Rebecca Childs. In preparation. "Narrative as Cognitive Artifact: The Case of Beowulf."
Hirschfeld, Lawrence A., and Susan A. Gelman. Mapping the Mind: Domain Specificity in Culture and Cognition. New York: Cambridge University Press.
Hutchins, Edwin. 1995. Cognition in the Wild. Cambridge, MA: MIT Press.
Jahn, Manfred. 1997. "Frames, Preferences, and the Reading of Third-Person Narratives: Towards a Cognitive Narratology." Poetics Today 18: 441-68.
-----. 1999. "'Speak, friend, and enter': Garden Paths, Artificial Intelligence, and Cognitive Narratology." In David Herman, ed., Narratologies: New Perspectives on Narrative Analysis, pp. 167-94. Columbus: Ohio State University Press.
James, William. 1890. Principles of Psychology. New York: Dover.
Kafalenos, Emma. 1997. "Functions after Propp: Words to Talk about How We Read Narrative." Poetics Today 18: 469-94.
Kahneman, Daniel, Paul Slovic, and Amos Tversky, eds. 1982. Judgment under Uncertainty: Heuristics and Biases. New York: Cambridge University Press.
Lévi-Strauss, Claude. 1986. "The Structural Study of Myth." In Hazard Adams and Leroy Searle, eds., Critical Theory since 1965, pp. 809-22. Tallahassee: University Presses of Florida.
Linde, Charlotte. 1993. Life Stories: The Creation of Coherence. Oxford: Oxford University Press.
Locke, John L. The Child's Path to Spoken Language. Cambridge, MA: Harvard University Press.
McKoon, Gail, Gregory Ward, Roger Ratcliff, and Richard Sproat. 1993. "Morphosyntactic and Pragmatic Factors Affecting the Accessiblity of Discourse Entities." Journal of Memory and Language 32: 56-75.
MacNeil, Lynda D. 1996. "Homo Inventans: The Evolution of Narrativity." Language & Communication 16.4: 331-60.
Mandler, Jean Matter. 1984. Stories, Scripts, and Scenes: Aspects of Schema Theory. Hillsdale, NJ: Lawrence Erlbaum Associates.
Margolin, Uri. 1986. "The Doer and the Deed: Action Basis for Characterization in Narrative." Poetics Today 7: 205-26.
-----. 1990. "Individuals in Narrative Worlds: An Ontological Perspective." Poetics Today 11: 843-71.
-----. 1999. "Of What Is Past, Is Passing, or to Come: Temporality, Aspectuality, Modality, and the Nature of Narrative." In David Herman, editor, Narratologies: New Perspectives on Narrative Analysis, pp. 142-66. Ohio State University Press.
Mateas, Michael, and Phoebe Sengers, eds. 1999. Narrative Intelligence: Papers from the 1999 AAAI Fall Symposium. Technical Report FS-99-01. Menlo Park, CA: American Association for Artificial Intelligence Press.
Minsky, Marvin. 1975. "A Framework for Representing Knowledge." The Psychology of Computer Vision. Ed. Patrick Winston. New York: McGraw-Hill, 1975. 211-77.
Norman, Donald. 1993. Things That Make Us Smart: Defending Human Attributes in the Age of the Machine. Reading, MA: Addison-Wesley.
Ochs, Elinor, Carolyn Taylor, Dina Rudolph, and Ruth Smith. 1992. "Storytelling as Theory-Building Activity." Discourse Processes 15: 37-72.
Palmer, Alan. "The Construction of Fictional Minds." Narrative (forthcoming).
Pavel, Thomas G. 1985. The Poetics of Plot: The Case of English Renaissance Drama. Minneapolis: University of Minnesota Press.
-----. 1986. Fictional Worlds. Cambridge: Harvard University Press.
-----. 1989. The Feud of Language: A History of Structuralist Thought. Trans. Linda Jordan and Thomas G. Pavel. Cambridge: Basil Blackwell.
Pinker, Steven. 1994. The Language Instinct. New York: Penguin Books.
Pratt, Mary Louise. 1977. Towards a Speech Act Theory of Literary Discourse. Bloomington: Indiana University Press.
Prince, Gerald. 1973. A Grammar of Stories. The Hague: Mouton.
Propp, Vladimir. 1968. Morphology of the Folktale. 2nd edition. Trans. Laurence Scott; revised by Louis A. Wagner. Austin: University of Texas Press.
Rimmon-Kenan, Shlomith. 1983. Narrative Fiction: Contemporary Poetics. London: Methuen.
-----. 1989. "How the Model Neglects the Medium: Linguistics, Language, and the Crisis of Narratology." Journal of Narrative Technique 19: 157-66.
Rummelhart, David E. 1980. "On Evaluating Story Grammars." Cognitive Science 4: 313-16.
Ryan, Marie-Laure. 1979. "Linguistic Models in Narratology: From Structuralism to Generative Semantics." Semiotica 28-1/2: 127-55.
-----. 1991. Possible Worlds, Artificial Intelligence, and Narrative Theory. Bloomington: Indiana University Press.
Schank, Roger C. and Robert P. Abelson. 1977. Scripts, Plans, Goals and Understanding: An Inquiry into Human Knowledge Structures. Hillsdale, NJ: Lawrence Erlbaum Associates.
Todorov, Tzvetan. 1968. "La Grammaire du récit." Langages 12: 94-102. -----. 1969. Grammaire du "Décaméron." The Hague: Mouton.
Tolman, Edward C. 1948. "Cognitive Maps in Rats and Man." Psychological Review 55: 189-208.
Turner, Mark. 1996. The Literary Mind. New York: Oxford University Press.
Tversky, Amos, and Daniel Kahneman. 1971. "Belief in the 'law of small numbers.'" Psychological Bulletin 76: 105-110.
-----. 1974. "Judgment under Uncertainty: Heuristics and Biases." Science 185: 1124-31.
Vygotsky, Lev. 1978. Mind in Society: The Development of Higher Psychological Processes. Ed. Michael Cole. Cambridge, MA: Harvard University Press.
Walker, Marilyn A., Aravind K. Joshi, and Ellen F. Prince, eds. 1998. Centering Theory in Discourse. Oxford: Oxford University Press.
Walton, Kendall. 1990. Mimesis as Make-Believe: On the Foundations of the Representational Arts. Cambridge, MA: Harvard University Press.
Webber, Bonnie Lynn. 1979. A Formal Approach to Discourse Anaphora. New York: Garland.
White, Hayden. 1996. "The Value of Narrativity in the Representation of Reality." In Susan Onega and José Angel García Landa, eds., Narratology: An Introduction, pp. 273-85. London: Longman.
Wilson, Robert A., and Frank C. Keil, eds. 1999. The MIT Encyclopedia of the Cognitive Sciences. Cambridge, MA: MIT Press.
Young, R. Michael, and David Herman. 2000. "Combining Artificial Intelligence and Narrative Theory: Action Representations for Interactive Learning Environments." Project funded by a Faculty Research and Professional Development Grant from North Carolina State University.
This site is optimized for Netscape 6 and higher
site design: Sara Roegiers @ Maerlantcentrum