View in EDS

Implications of Human Pattern Processing for the Design of Artificial Knowledge Systems.

Saved in:
Bibliographic Details
Title: Implications of Human Pattern Processing for the Design of Artificial Knowledge Systems.
Language: English
Authors: Hayes-Roth, Barbara, Rand Corp., Santa Monica, CA.
Availability: Rand Corporation, Main St., Santa Monica, CA 90406 ($3.00).
Peer Reviewed: N
Page Count: 31
Publication Date: 1977
Document Type: Information Analyses
Descriptors: Artificial Intelligence, Cognitive Processes, Cues, Cybernetics, Design Requirements, Epistemology, Literature Reviews, Pattern Recognition
Abstract: This paper presents evidence that four design principles commonly embodied in artificial knowledge systems are inconsistent with human cognitive capabilities. Because these principles are widely accepted as characteristics of human knowledge processing, common theoretical properties related to cognitive psychology and artificial intelligence which are apparent in pattern processing research, are considered, including (1) canonical meaning representations of stored patterns; (2) prototypes as representations of classes of stored patterns; (3) pattern-matching processes that must fully evaluate all stored patterns that partially match a target pattern; and (4) pattern-matching processes that are influenced by pattern complexity and limited processing capacity. This characterization of human pattern processing is critically evaluated in light of recent psychological studies, and the implications of these results for artificial knowledge systems design are discussed. Forty-two references are listed. (LMM)
Entry Date: 1983
Accession Number: ED230186
Database: ERIC
Description
Abstract:This paper presents evidence that four design principles commonly embodied in artificial knowledge systems are inconsistent with human cognitive capabilities. Because these principles are widely accepted as characteristics of human knowledge processing, common theoretical properties related to cognitive psychology and artificial intelligence which are apparent in pattern processing research, are considered, including (1) canonical meaning representations of stored patterns; (2) prototypes as representations of classes of stored patterns; (3) pattern-matching processes that must fully evaluate all stored patterns that partially match a target pattern; and (4) pattern-matching processes that are influenced by pattern complexity and limited processing capacity. This characterization of human pattern processing is critically evaluated in light of recent psychological studies, and the implications of these results for artificial knowledge systems design are discussed. Forty-two references are listed. (LMM)