ABSTRACTThe present study focused on the effects of, and the interactions between, practice and task structure on human performance. The development of automatic processing through consistent stimulus‐response mapping (CM) was assessed by means of measures of reaction time and event‐related brain potentials. The subjects performed a visual search task in which they responded by pressing a button whenever a probe matched a memory set item. The variables manipulated in the study included the number of memory set items (1 or 4), the task structure (CM or VM), and the probability of occurrence of a memory set item (.2 or .8). Set size had a significant effect on RT in both CM and VM conditions prior to practice and in the VM condition following extensive practice. P300 latency mirrored RT, suggesting that the development of automatic processing substantially reduced stimulus evaluation time. The commonly observed relationship between probability and P300 amplitude, with larger P300s elicited by infrequent events, was found in the VM conditions but not in the CM condition after practice. Two different negative components were affected by stimulus mismatch. The N200 component behaved in a similar manner during automatic and controlled processing. However, a late frontally negative component was sensitive to the degree of mismatch only during controlled process