• Choose a category
• All
• Classic-Fiction

By Root 600 0

I find the results of such an application both exciting and unnerving. On the one hand, I can argue now with a reasonable degree of confidence that certain aspects of Woolf's prose do place extraordinarily high demands on our mind-reading ability and that this could account, at least in part, for the fact that many readers feel challenged by that novel. On the other hand, I came to be "afraid" of Mrs. Dalloway— and, indeed, other novels—in a different fashion, realizing that any initial inquiry into the ways fiction teases our ToM immediately raises more questions about ToM and fiction than we are currently able to answer. My ambivalence, in other words, stems from the realization that ToM underlies our interaction with literary texts in such profound and complex ways that any endeavor to isolate one particular aspect of such an interaction feels like carving the text at joints that are fundamentally, paradigmatically absent.

This proviso should be kept in mind as we turn to the experiments investigating one particular aspect of ToM, namely, our ability to navigate multiple levels of intentionality present in a narrative. Although ToM is formally defined as a second-order intentionality—for example, "I believe that you desired" or Peter Walsh thinks that Clarissa "would think [him] a failure" (43)—the levels of intentionality can "recurse" further back, for example, to the third level, as in the title of George Butte's wonderful recent book, I Know That You Know That I Know" or to the fourth level, as in "I believe that you think that she believes that he thinks that X," and so forth. Dennett, who first discussed this recursiveness of the levels of intentionality in 1983, thought that it could be, in principle, infinite. A recent series of striking experiments reported by Dunbar and his colleagues have suggested, however, that our cognitive architecture may discourage the proliferation of cultural narratives that involve "infinite" levels of intentionality.

In those experiments, subjects were given two types of stories. One cluster of stories involved a "simple account of a sequence of events in which A gave rise to B, which resulted in C, which in turn caused D, etc.'" Another cluster introduced "short vignettes on everyday experiences (someone wanting to date another person, someone wanting to persuade her boss to award a pay rise), . . . [all of which] contained between three and five levels of embedded intentionality." Subjects were then asked to complete a "series of questions graded by the levels of intentionality present in the story," including some factual questions "designed to check that any failures of intentionality questions were not simply due to failure to remember the material facts of the story." The results of the study were revealing: "Subjects had little problem with the factual causal reasoning story: error rates were approximately 5% across six levels of causal sequencing. Error rates on the mind-reading tasks were similar (5-10%) up to and including fourth-level intentionality, but rose dramatically to nearly 60% on fifth-order tasks." Cognitive scientists knew that this "failure on the mind-reading tasks [was] not simply a consequence of forget

7: Cognitive Science and Mrs. Dalloway

ting what happened, because subjects performed well on the memory-forfacts tasks embedded into the mind-reading questions."' The results thus suggest that people have marked difficulties processing stories that involve mind-reading above the fourth level.2

An important point that should not be lost in the discussion of the experiments reported by Dunbar is that it is the content of the information in question that makes the navigation of multiply embedded data either relatively easy or relatively difficult. Cognitive evolutionary psychologists suggest the following reason for the ease with which we can process long sequences, such as, "A gave rise to B, which resulted in C, which in turn caused D, which led to E, which made possible F, which eventually brought about G, etc.," as opposed to similarly long sequences