25 experiments on semantic generalization of classically conditioned responses are reviewed. Several major methodological inadequacies are discussed. These included frequent failure to convincingly demonstrate conditioning, failure to distinguish between generalization effects and extinction effects, and failure to control for simultaneous conditioning to the generalization stimuli. 2 alternative inferred mechanisms of generalization are discussed. While mediation by common responses is not strongly supported by evidence, there is suggestive evidence of the importance of S's categorizing generalization stimuli as similar to the conditioned stimuli, as evidenced by Ss' reported expectation of reinforcement.

No abstract