Comments on Meir Sternberg's "Perspectives on Three Decades of Cognitivist Efforts"
Art Graesser
University of MemphisIntroduction | Gerald Cupchik | Andrew Elfenbein | | Uri Margolin | Alan Richardson | Meir Sternberg
I read with fascination Meir Sternberg's perspectives on some of the efforts of researchers in discourse psychology, artificial intelligence, cognitive science, and discourse processing over the last 30 years. As one who has built a career in interdisciplinary studies, I am always fascinated by the views of researchers in other fields when they take a peek at what we do in our shops. I was impressed with the scholarly coverage and the care with which Sternberg analyzed our scientific insights and activities. I was flattered to be among the targets in his bewildering black list of ad hominums, but rather envious that Bill Brewer was pummeled far more than I was. I was tantalized by prose that was dense with scholarly nuances and that challenged the working memories of virtually all of us readers. This is indeed scholarship at its finest. At the same time, I was also amused as I read what a person in narratology has to say about a scientific enterprise. Sternberg is indeed a consummate narratologist and literary scholar, but mighty green behind the ears when it comes to understanding the goals and practice of science.
I entirely agree with Sternberg that there needs to be a deeper and more detailed integration of narratology and literary studies with cognitivist efforts. I also have worried about this lack of deep integration. A few years ago I wrote a column in an IGEL newsletter proposing that researchers from different fields need to sleep together before they achieve a deep enough integration. (For the record, this was an entirely metaphorical plea, not necessary a literal one.) Sternberg's scholarly contribution will be valuable if this message would get translated into practice. Unfortunately, there was no clue in his arguments how to achieve a deeper cross-fertilization. So let me now turn to substantive criticism of Sternberg.
(1) Sternberg does not appreciate the fact that the scientific method encourages simplification of basic mechanisms, rather than an exhaustive inventory of potential factors that affect a phenomenon. Where is the evidence that cognitivists missed the boat on basic cognitive mechanisms?
(2) Discourse psychologists have investigated the dynamic process of constructing narrative interpretations on-line, including expectations of future occurrences, reinterpretations of prior constituents, inferences that connect text segments, time frame shifts, shifts in character and spatial perspectives, thematic structures, affect of characters and readers, intertextuality, and other aspects of narrative that are well beyond the arena of story grammars. Dozens of researchers have collected think aloud protocols, eye tracking data, reading times, and word priming data while text is read, sentence by sentence, in an effort to track the dynamic processes during reading as the construction of a narrative microworld unfolds. Discourse psychology of 2004 is not the same as in 1984. Sternberg is fixated on the early days of cognitivism, when the field just got started, rather than the cognitivism of today.
(3) Sternberg does not understand the capabilities and goals of computer simulations of comprehension. Computer models of comprehension are progressively becoming more sophisticated each year. There are now computer systems that grade essays as well as experts in English composition. There are computer systems that hold captivating conversations with people and others that have surprising capabilities of speech recognition. The field is blossoming from a revolution of advances in corpus and computational linguistics. These systems are based on architectures with multilayered, probabilistic, soft constraint satisfaction. Searle's 20-year old Chinese room argument about systems with brittle rules is now in the jaded annals of history - no longer relevant in contemporary cognitive science. Aside from the successes of such computer models, one important theoretical goal of computer simulation is to discover how theories are incomplete or inconsistent, not to assert that a model perfectly mimics human cognition. Consequently, Sternberg's observations about the limitations of such systems have little impact.
(4) Sternberg criticizes Brewer's structural affect theory that investigated suspense, surprise, and curiosity in narrative. Brewer's theory was a pioneering advance in discourse psychology in the early 1980's, but undoubtedly incomplete -- as early scientific theories often are. Aside from the validity of Brewer's contribution as a psychological theory, Sternberg does not offer a shred of evidence that it is flawed from the standpoint of psychological data or a corpus analysis of relevant narratives.
In closing, it is exciting that Sternberg's thoughtful scholarly contribution has fanned the flames of debate in our field. I am entirely convinced that our world of IGEL will only be energized by the mighty, challenging sword of Sternberg. I can't wait to see where the debate and aftermath will evolve.
Document created January 26th 2004