Interaction of Human Cognitive Models and Computer-Based Models in Supervisory Control
Abstract
This report summarizes the first year's effort of a three-year research project on how knowledge is represented in decision aids and control systems and how the operators of such systems apparently represent and utilize such knowledge. The first section of the report discusses the relationship of computer-based supervisory control to computer-based decision-aiding (expert systems) by identifying component variables and functions and building up block diagrams. The second section deals quantitatively with internal models, knowledge, and calibration, both with respect to expectations of the existence of identifiable states of the world and with respect to the overlap of meanings of terms (mental or linguistic encodings, fuzzy variables). The third section discusses mental models and their importance in three kinds of activities supervisors must do in complex systems: (1) discovering how things work; (2) determining what is wanted out of the set of alternatives states of the attributes; (3) encoding and manipulating fuzzy concepts; (4) combining evidence and confidence; (5) deciding what to do. The fourth section of the report deals with the human use of computer-based models in automatic control and in decision-aiding. It reports on three sets of experiments underway or completed.
Document Details
- Document Type
- Technical Report
- Publication Date
- Mar 01, 1984
- Accession Number
- ADA142547
Entities
People
- T. B. Sheridan
Organizations
- Massachusetts Institute of Technology