Cognitive genre theory I: Intro & writing as situated problem-solving

18 Mar

I’ve been posting the answers I’ve given to the written prelim exams for my PhD. (My specialty exam answer appears here.) Note that I have not provided the examiners’ questions. That’s because it’s considered impolite to share an examiner’s question without his/her permission. Anyway, here is my answer to the rhetoric question I received. Please be charitable if you are reading these posts. I’m only lightly editing them as I cut them from the exam I wrote in a hurried 24 hours.

This series of posts explores the possibility of a cognitive theory of genre. It does so tentatively and preliminarily, recognizing the (comparatively) limited time I have considered the issues and the (comparatively) small list of texts upon which I base my assessments of the state of the art in contemporary genre theory. This first post explores several theories in cognitive science, and the second discusses a movement in pragmatics, both of which have had little or no uptake in genre theory. The third and fourth posts propose the initial steps for formulating a cognitive model of genre theory. I’ll attempt to relate these poses to the previous work in genre theory, with special attention to Miller, Swales, Bhatia, and Berkenkotter & Huckin.

In 2004, Amy Devitt conceived of the role of cognitive research in the context of genre studies as extending the arguably process- and product-focused research from the 1980s by Linda Flower and John Hayes. Though much-praised in its time and by many since then for its quality and thoughtfulness, the research of Flower and Hayes has also been criticized on grounds that it “lacks a crucial social dimension” (Swales 1990, p.4). Though the Flower and Hayes research and other research from what has been called the “influential Carnegie Mellon model” (ibid.) is not on the present reading list, it casts a shadow over the role of cognitive methods in writing research today. Before offering a cognitive theory of genre, then, it will be helpful to offer some observations about contemporary cognitive science, its departure from the Carnegie-Mellon model of the early 1980s, and its uptake (or lack thereof) in the genre literature.

The balance of this post addresses contemporary approaches to problem-solving associated with situated cognition theories, contrasting them with the classical model of Carnegie-Mellon fame. Flower and Hayes were influenced by the work of Herbert Simon—polymath and Nobel prize winner—and Allen Newell—computer/cognitive science researcher and Turing Award winner—both of whom were active at Carnegie-Mellon in developing what is now called the “classical theory” of problem-solving. Kirsh (2009) recounts some characteristics of the classical model as articulated by Newell and Simon. That model contemplates a task environment, the “abstract structure that corresponds to the problem” (p. 265). The task environment includes those aspects of the material environment that are relevant to solving the problem (like the chess pieces if the problem is chess) but not those aspects deemed irrelevant to the problem (like color of the walls of the room where the chess game is played). The problem-solving itself is conceived of as search through a problem space, the mental representation encoded by an agent (but possibly included in external forms) who understands a task correctly, and consists of the following:

  • An initial state.
  • “A representation of the goal state or condition” or a test for determining whether it has been reached (p. 266).
  • “Constraints determining allowable moves and states” (p. 266).
  • Other useful representations, including “problem-solving methods, heuristics, or metrics specific to the current task environment” (p. 266).

Note the use of the term “agent” here; it is commonly used in cognitive science. Whether one argues that cognition happens in the brain or the central nervous system (Searle, 2000), in the whole body (Rowlands, 2010), in the body and its material environment (Kirsh, 2009), or in the body and its material and social environments (Clark, 1997), from a phenomenological perspective, individual human beings perceive themselves as individuals, as thinking, acting beings; most believe they have the power to act. I adopt the term “agent” then to refer to an individual in a situation where she might act (or choose not to act) in order to advance some intended purpose.

The classical model was developed based on experiments with well defined problems (like playing checkers), those where the initial state, goal state, and constraints are capable of formal description. Even classical theorists acknowledge that most problems are not well defined—but rather are ill-defined—Kirsh says that an ill-defined problem is “largely being made up as it is being worked on” (p. 268).

The classical model sees problem-solving as search through the problem space, but before the search it calls for two preliminary steps: framing and registration. Framing is deciding “what is relevant and what is irrelevant” (p. 268). This is no great problem in “knowledge lean” tasks, such as games and puzzles, where the initial and goal states and rules can be stated exhaustively. But Kirsh claims that in everyday solving of ill-defined problems, it is the “setting and local resources” that “activate an interpretive framework that primes agents to look for and conceptualize features of their environment in activity-specific ways” (p. 271). He gives the example of shoppers in supermarkets ignoring unit prices printed on product displays and instead performing an informal mathematical analysis that takes into account real-world considerations like where the product will be stored and whether it will be used before it spoils. The classical model also requires registration, responsible for connecting the “abstract [problem] search space… to the real world and then reinterpreting the results of the search” (p. 269). Kirsh argues this, too, cannot be an effort restricted to the beginning and end of problem-solving, but that with most problems “the search process is complicated by the need to continually anchor the search space in locally meaningful ways” (p. 276). He offers the example of using a paper map to find a route: Usually a map user does not view the map, formulate a route-plan, and then execute it without further reference to the map. Instead the agent will use physical landmarks to confirm location and orientation. Then, as the agent follows the route, s/he continues to confirm that physical landmarks correspond to the map. In well-defined problems, it might be possible to view registration as distinct from searching for the solution, but in everyday (mostly ill-defined) problems, Kirsh argues that situatedness of problem solving takes the agent back and forth between mental representations and interactions with the environment.

This leads Kirsh to a discussion of interactivity and “epistemic actions.” He claims the classical model did/does not account for the interactivity described above, because it assumes “that users completely search an internal representation of their problem before acting” and fails to see interactivity as “a force for reshaping either the search process or the problem space” (p. 278). Kirsh describes epistemic actions as ones that agents perform to learn about the problem environment, verify desirability of certain states, and test plans. For example, event expert players of the video game Tetris rotate game “pieces” as they are falling to learn about their possibilities. The classical theory does not account for these actions as it has no category for them. Kirsh’s theory considers resources and scaffolds, “cultural products and artifacts” (p. 284), that play a role in problem-solving because our environments are constructed to be helpful for the activities we wish to carry on in them. He uses the example of constructing a geometric drawing to solve a problem in analytical geometry; this transformation of the problem may allow the agent to find a solution more readily. Philosopher of mind Andy Clark (1997) also contemplates various social practices among the “scaffolding” agents use in carrying out cognitive activities.

(An aside about the use of “scaffolding” in this context: Kirsh references Vygotsky for the scaffolding concept, where a teacher or expert helps a learner or novice by offering tips, tools, heuristics and the like when the learner or novice is at the “zone of proximal development”—that is, at the furthest extent of his/her current capabilities. The teacher then removes the scaffolding after the student moves on. Note that some cognitive scientists and philosophers of mind have adapted Vygotsky’s concept to refer to the matrix or structure of a culture in its role as a support for cognitive activity. (See, e.g., Clark, 1997.) This type of “scaffolding” is not removed after the learner has moved beyond the “zone of proximal development”; instead, it remains as part of the learner’s cultural resources.)

So, we’ve seen how some contemporary conceptions of cognitive science have moved away from the isolated problem as the locus of investigation and instead recognizes the complex environments in which ill-defined problems have their existence. In the next post, we’ll look at two concepts from cognitive science that pre-date the social turn in genre theory, but which nevertheless have not received deep attention in genre theory.

-Brian

Read the next post

Advertisements
%d bloggers like this: