Toward Integrated Cognitive Systems, Which Must Make Fuzzy Decisions About Fuzzy Problems

Abstract

This paper presents and gives examples of the behavior of a simple computer-programmed model (called "SEER", for - semantic learner - ), for an integrated, wholistic cognitive system. The system combines the cognitive functions of pattern recognition, scene description, information retrieval, deducing, and acting that are usually handled in separate programs. This forces it to make fuzzy decisions, e.g. as to what type of behavior (e.g. describe or answer) to effect, and what type of thing (e.g. word or object) is being perceived. The paper examines a number of aspects of such a system, in order to illustrate how it begins to handle the problems of fuzziness that can no longer be avoided, and should no longer be avoided, since they appear to be at the heart of intelligence.