Procedural learning: II. Intertrial repetition effects in speeded-choice tasks.

Previous research has demonstrated that subjects' responses in a speeded-choice task are affected by the sequence of stimuli and responses. Generally, responses are faster when the same stimulus is repeated. There is disagreement in the literature concerning whether or not this effect represents a speedup of perception, response execution, or central stages of processing. The first 5 experiments reported here demonstrate that the effect is quite stimulus specific: A succession of different stimuli that call for the same response produces little repetition benefit, unless the stimuli differ only in very superficial ways (e.g., in color). Increasing the intertrial interval from 100 ms to 1000 ms attenuates these repetition effects only slightly. These results would be consistent with a perceptual locus for the repetition effect, but Experiment 6 shows that if the response mapping changes from trial to trial, the advantage for stimulus repetitions is abolished The results indicate that locus of the repetition effect is at the stage of response selection. However, the stimulus specificity of the effects indicates that immediate repetitions produce or strengthen transient links that "shortcut" the response-selection stage. By contrast, the results of Pashler and Baylis (1991) indicate that practice primarily strengthens response selection at the categorical level rather than shortcutting it One implication of the results is that the practice effect is not created simply by the accumulation of the traces responsible for repetition effects.