An evaluation of the total similarity principle: effects of similarity on frequency judgments

Memory models that embody the total similarity principle (e.g., Gillund & Shiffrin, 1984; Hintzman, 1988; Murdock, 1982; Ratcliff, 1990) assume that frequency judgments reflect the total similarity of a test item to stimuli that have been studied. In 4 experiments, subjects estimated the frequencies of target words that had been presented in the context of varying numbers of semantically similar words. In a fifth experiment, subjects made forced-choice relative frequency judgments. The results of these experiments supported 1 prediction of total similarity models: Presenting similar words will increase rather than decrease frequency judgments of target words. However, a second prediction of these models was not supported. In particular, similar-word presentations had no effect on the judged frequencies of target words that had not been shown.