Output interference in the recall of categorized and paired-associate lists.

Abstract

Three experiments are reported in which, following presentation of a categorized list, subjects attempted recall of words within categories when given the category names as cues. Recall of words declined with the test position of the category. This output interference was not increased (a) when some items from each category were given as cues in addition to the category name (Experiment 1), (b) when the first categories recalled contained nine rather than three words (Experiment 2), and (c) when the categories in the list were semantically related rather than unrelated (Experiment 3). Interference in recall of categories tested later was not affected by the number of words recalled from prior categories. Output interference in paired-associ ate recall was obtained in a fourth experiment. These experiments indicate that the act of recall can itself serve as a source of forgetting. Implications of the results for several theories of recall are discussed. Several different lines of evidence have been used to show that one source of forgetting is the act of recall itself. (See Roediger, 1974, for a review.) The first systematic study of output interference was that of Tulving and Arbuckle (1963). They presented subjects with lists of 10 paired associates in which the digits 1-10 were the stimuli and common words were the responses. Subjects' recall was cued by the stimuli, with output order counterbalanced so that each stimulus occurred equally often at each output position. The primary finding was that recall of responses that had been presented in the initial serial positions of the input list did not vary with output position, but recall of responses presented late in the input sequence became increasingly poor across the output positions. Output interference was most dramatic for the last two or three items, but there was some evidence