ABSTRACT
Sequential effects in loudness judgments were studied in a four-stimulus absolute judgment task, a two-stimulus × two-response identification task, and a two-stimulus task with confidence ratings. Multiple regression analyses of the data showed that the current stimulus, previous stimulus, and previous response influenced the choice Of the current response, which was contrasted with the previous stimulus and assimilated to the previous response in all three tasks. It was proposed that subjects judge the difference between the current and previous stimuli rather than the current stimulus itself and shift their response criteria so as to reduce biases created in the sensory evidence by the shifting value of the trial n - 1 comparison stimulus. Reaction-time data showing that responses were slower on trials when the current stimulus was a repetition of the previous one also supported the hypothesis that judgments involve sampling from the distribution of differences between the trial n - 1 and trial n stimuli to choose a response.
