Related Experiment Videos
Assessment of the Rescorla-Wagner model
R R Miller1, R C Barnet, N J Grahame
1Department of Psychology, State University of New York at Binghamton 13902-6000, USA.
Psychological Bulletin
|May 1, 1995
Summary
The Rescorla-Wagner model is a key theory in associative learning. While influential, its successes and failures highlight limitations, suggesting it
Area of Science:
- Behavioral Psychology
- Computational Neuroscience
- Machine Learning
Background:
- The Rescorla-Wagner model is a dominant theory in associative learning.
- Connectionist models are often benchmarked against the Rescorla-Wagner model.
- The Widrow-Hoff delta rule is frequently used to achieve model equivalence.
Purpose of the Study:
- To present the Rescorla-Wagner model's core assumptions.
- To review the model's predictive successes and failures.
- To discuss the model's heuristic value and limitations.
Main Methods:
- Review of Rescorla-Wagner model assumptions.
- Analysis of model's predictive accuracy.
- Discussion of model's influence on research and development.
Main Results:
- The Rescorla-Wagner model has demonstrated significant predictive successes.
- Specific failures of the model are linked to its underlying assumptions.
- The model has stimulated research and new model development in associative learning.
Conclusions:
- The Rescorla-Wagner model has positively influenced simple associative learning research.
- The model should not be considered definitively "correct".
- Its predictive accuracy is not a universal benchmark for assessing other models.