The Effects of Contextual Changes on Component Recognition

Abstract

Triplets of verbal items were followed by tests for recognition of one of the items, the other two items (context) being intact, permuted, removed from display, or replaced by new items. Recognition was best when the test context was identical to that during training. The data seem to reflect specific retrieval processes, especially the generating of codes from the test item and the comparing of those codes against multiple representations in memory. A fundamental variable in the retrieval of items from memory is the completeness with which the original pattern of stimulation is presented during the tests. McGeoch (1942) has reviewed a number of experiments which demonstrate that recall performance clearly depends on the degree to which the training stimuli are reinstated on the tests. Similarly, experiments involving alterations in 'incidental' background stimuli have produced substantial reductions in the recall and relearning of paired associates (Peterson and Peterson, 1957; Weiss and Margoulis, 1954). Moreover, experiments using a freerecall paradigm have shown that the introduction of 'cue' words during the test phase may either facilitate or inhibit the recall of individual words (Postman, Adams, and Phillips, 1955; Tulving and Pearlstone, 1966; Bower, Clark, Lesgold, and Winzenz, 1969). Although the evidence about the effects of altered test conditions on recall and relearning is quite substantial, experimentation with verbal materials and other measures of retention, particularly recognition, has only recently been reported (Light and Carter-Sobell, 1970; Tulving and Thomson, 1971). The present experiment was an attempt to extend these recent findings by examining how contextual changes might produce variations in the correct recognition of verbal 'units' previously presented in a compound display.