A Review of Production System Models of Cognition and Example Demonstration

BRANNON, NATHAN G.

Sandia National Laboratories

关键词: 99 General And Miscellaneous//Mathematics, Computing, And Information Science;    Processing;    Brain;    Learning;    Man;   

DOI  :  10.2172/808592
RP-ID  :  SAND2002-3727
RP-ID  :  AC04-94AL85000
RP-ID  :  808592

美国|英语

来源: UNT Digital Library

【 摘 要 】

There have been significant efforts to develop cognitively plausible software architectures of human information processing in the last three decades. This report summarizes several architectures that continue to be developed. The specific type of cognitive models developed are known as production system architectures, which refers to the characterization of knowledge in terms of procedural (''how-to'' knowledge) condition-action relationships consisting of declarative (''what'' or factual) knowledge. To illustrate the ability for these models to instantiate human cognitive performance, a simulation using ACT-R (Adaptive Control of Thought - Rational) was implemented for a supervisory control task. Correlations between simulated and human learning of the task were measured and yielded correlations as high as 0.93.