Foundations Of Cognitive Science

Rescorla-Wagner Model

The Rescorla-Wagner models is a model of associative learning that was proposed in the early 1970s to account for a variety of complex regularities concerning associative learning (Rescorla & Wagner, 1972). In general terms, it defines the change in associative strengths between stimuli and responses in terms of the difference between existing strengths (for all cues present) and the maximum strength possible. In other words, it is a "cognitive" theory of associative learning, in the sense that learning occurs when "expectations" are violated.

The Rescorla-Wagner model is of interest to cognitive science not only because of its power and importance (Miller et al., 1995), but because there is a formal equivalence between this learning rule and the delta rule used to train artificial neural networks (Sutton & Barto, 1981). Importantly, such networks can generate different responses than those predicted by the Rescorla-Wagner model, which can be explained by the fact that the Rescorla-Wagner model changes associations in the absence of behavior, while neural networks cannot (Dawson, 2008).


  1. Dawson, M. R. W. (2008). Connectionism and classical conditioning. Comparative Cognition and Behavior Reviews, 3 (Monograph), 1-115.
  2. Miller, R. R., Barnet, R. C., & Grahame, N. J. (1995). Assessment of the Rescorla-Wagner model. Psychological Bulletin, 117(3), 363-386.
  3. Rescorla, R. A., & Wagner, A. R. (1972). A theory of Pavlovian conditioning: Variations in the effectiveness of reinforcement and nonreinforcement. In A. H. Black & W. F. Prokasy (Eds.), Classical Conditioning II: Current Research And Theory (pp. 64-99). New York, NY: Appleton-Century-Crofts.
  4. Sutton, R. S., & Barto, A. G. (1981). Toward a modern theory of adaptive networks: Expectation and prediction. Psychological Review, 88(2), 135-170.

(Added March 2010)