Retrieval processes in continuous recognition.

The accuracy and response latency of yes/no recognition decisions were measured in three experiments by the continuous recognition paradigm. The principal independent variable was lag, or the number of intervening items between target presentations. Lag was varied from 0 to 40. A logarithmic function provided a good description of the relation between lag and correct response latency. Item repetition affected the intercept of the logarithmic functions, with little effect on the slope. A noun/nonnoun stimulus manipulation affected the slope of the functions with no appreciable effect on the intercept. The latter result was obtained both for once-repeated and twice-repeated item functions, and both when the stimulus manipulation was a between-lists variable (Experiment 2) and a within-lists variable (Experiment 3). The results were found to be incompatible with Atkinson and Juola's (1973) model and Murdock's (1974) conveyor-belt model. The retrieval theory of Ratcliff (1978) and the multiple-observations model of Pike, Dalgleish, and Wright (1977) provide the most satisfactory account of the present results. However, both models may have difficulty in accounting for the obtained repetition effects.