Serial recall and recognition in short-term memory tasks, interactions with test expectancy in the pre-cued post-cued paradigm

grantor: University of TorontoRecently, process models have been proposed to account for memory performance in serial recall and recognition tasks. Explanations of serial recall and recognition in the models are based on assumptions of how item and order information are encoded and retrieved in a single memory system. To date, the data used to constrain encoding assumptions of item and order information have come from experiments in which subjects are always pre-cued for either serial recall or recognition. Because pre-cued subjects are informed of the type of test before list presentation, they can use knowledge of the ensuing test to define what characteristics of the study list are relevant for encoding. Thus, there is a tacit, but crucial, assumption in the models that encoding is independent of expectation for the kind of test received. Because the models depend on encoding assumptions to predict serial position effects, it is necessary to empirically test whether expecting, for example, serial recall will have any affect on recognition performance (and vice versa). A series of experiments was conducted in which subjects were led to expect either a serial recall test or a recognition test, but received the other test instead. The results showed that encoding of order information is dependent on whether subjects expected a serial recall test. For recognition, subjects expecting serial recall produced higher accuracy but longer latencies compared to subjects expecting recognition. The discrepancy between accuracy and latency for subjects expecting serial recall is inconsistent with explanations of recognition based on a direct mapping of encoded strength to familiarity. To account for these findings, a revision of one model, TODAM, was proposed. Another model, OSCAR was shown to be incompatible with both the serial recall and recognition data. The results support two main conclusions. First, contrary to the independent encoding assumption, the encoding of item and order information is dependent on task expectation. Second, the effect of test expectancy occurs at both the encoding and retrieval stages as opposed to only affecting the encoding stage.Ph.D