A computational theory of learning causal relationships (1991)
Cached
Download Links
- [www.ics.uci.edu]
- [www.cs.utexas.edu]
- [www.cs.utexas.edu]
- DBLP
Other Repositories/Bibliography
| Venue: | Cognitive Science |
| Citations: | 13 - 1 self |
BibTeX
@ARTICLE{Pazzani91acomputational,
author = {Michael Pazzani},
title = {A computational theory of learning causal relationships},
journal = {Cognitive Science},
year = {1991},
volume = {15},
pages = {401--424}
}
Years of Citing Articles
OpenURL
Abstract
I present D cognitive model of the humon ability lo acquire c.us.I relotionshipr. I report on experimental evidence demonrtroting that human leornerr acquire occwote cwxd relationships more rapidly when training examples oreconrirtent with o general theory of cwsolity. This article describes o learning procerr that uses o general theory of causality OS background knowledge. The leorning pro-cess, which I cdl theory-driven learning (TDL), hypothesizes cw~a1 relationships consistent both with observed doto and the general theory of courolity. TDL accounts for data on both the rote a+ which humon learners acquire couscll relo-tionrhips, and the types of COUSJ relationships they acquire. Experiments with TDL demonrtrote the odvontoge of TDL for acquiring cowa relationships over similarity-bored opproacher to learning: Fewer examples ore required to loom an acc~rote relotionrhio. 1.







