| Abstract: | The model of the decision system in Murdock's (1982, 1983) two-stage memory-and-decision model for item recognition is developed and tested. The underlying strength distributions are assumed to result from the operation of a distributed-memory system. The decision model assumes that extraneous noise is added to the result of the memory comparison process, and a decision is made when the total evidence falls below a lower criterion or rises above an upper criterion. The decision model is shown to be able to fit the accuracy and mean response latency data from four major recognition paradigms (Sternberg, study-test, continuous, and prememorized list). In addition, the decision model was also able to fit the response time distributions derived from the convolution analysis of Ratcliff and Murdock (1976), and the changes in the parameters of the distributions over experimental conditions in each paradigm. The model was also applied to speed-accuracy trade-off, repeated negatives, and the mirror effect. (PsycINFO Database Record (c) 2010 APA, all rights reserved) |
