Test list effects and the role of the conditioned response in semantic generalization

This study attempted to determine the locus of semantic generalization, whether generalization occurred during testing as well as training in a verbal learning task. The role of a conditioned response was examined. An attempt was made to replicate the results of a previous experiment, demonstrating a gradient of semantic generalization as a function of the associative strength of the verbal stimuli. 72 Ss were visually presented with a test list consisting of 144 words, incorporating six word categories. There were 24 critical stimulus (CS) words, and four associatively-related words (GSs) plus one control (X) word for each CS word. Ss responded by pressing buttons marked 1 to 6, according to their confidence during testing that the words had been presented during training. 36 Ss were presented with CS words during training; 36 Ss were presented with Clang words, i.e., words that were phonetographically similar to CS words. Half of the Ss in each of these groups were required to actively respond during training; half were not. A gradient of semantic generalization obtained across all conditions. Generalization occurred for Ss trained with Clang words as well as those trained with CS words, suggesting that some generalization does occur during testing. Ss not required to make a response during training responded as predicted to generalization stimuli during testing, indicating that a conditioned response is not necessary for generalization to occur.