Reminding and Memory Organization: An Introduction to MOPs
Roger C. Schank Yale University
A key question for researchers in cognitive science is the problem of how human memory is organized. The solution to this problem is fundamental to cognitive scientists regardless of whether their basic orientation is towards Artificial Intelligence or cognitive psychology. The theories that we test by means of psychological experiments ought to have some ramifications on how we build computer models of the processes tested by those experiments, and the computer models that we build ought to supply testable hypotheses for psychologists. This chapter discusses some issues in memory organization that are fundamental to an understanding system, and thus to cognitive science.
The continual evolution of our programs that attempt to understand natural language has recently begun to cause an interchange to occur between our group at Yale and various cognitive psychologists. In particular, our work on scripts, as embodied in the SAM ( Cullingford, 1978; Schank & Yale, 1975) and FRUMP ( DeJong, 1979) programs has caused various researchers outside of our group to attempt to test such notions experimentally ( Bower, Black, Turner, 1979; Graesser, Gordon, & Sawyer, 1979; Owens, Bower, & Black, 1979; Smith, Adams, & Schorr, 1978). Some of those experiments have begun to have an effect upon our theories and upon our subsequent programs. Much of this chapter, therefore, attempts to outline the problems we are trying to solve. The solutions we are proposing are at least partially a result of this interplay of psychological and computational concerns.
In addition to showing that script-like considerations are relevant in story understanding, one of the most valuable things to come out of empirical research on scripts was the problem it presented for us. Recognition confusions were