I also had another reaction: You have made me
realize that there is a fine line between "bad" recognizing, of the type
Phil describes, and "good" or "necessary" recognizing, as follows: All
social science theory involves generalizing/abstracting patterns (whether
"underlying," "prototypical," "deep structure) as descriptions for the
(widest possible) range of observed phenomena. Someone once said to me
that "everybody is a structuralist" in this sense.
I also often think about ways to generalize culturally shared patterns of
interaction and behavior, when I am not thinking about
praxis/improvisation/embodiment. Closely related to the "script" notion
is conversation analysis's "routine" concept, and I particularly like
Bill Corsaro's discussions of "routine embellishment" in practice.
But the general issue is: Can we really draw any line between the "bad"
examples of AI (from Phil) and our own theorizing about patterns and
structures? Phil, what do you think the difference is?
Keith Sawyer
Dept. of Psychology
UC Santa Cruz
Santa Cruz, CA 95064