We describe a protocol to examine various hypothetical models of task switching. This protocol analyzes two event-related potentials--lateralized readiness potential (LRP) and P300--to infer the roles of advance reconfiguration and carry-over effect on task switching. Participants performed two tasks in a random order. On each trial a task from the previous trial would be repeated, or the other task would be carried out. In one scenario, each stimulus was preceded either by an informative cue specifying which of the two tasks to perform (task-cueing conditions) or by a non-informative cue (no task-cueing conditions). The results showed that the mean reaction time and the stimulus-locked LRP intervals were longer for switch than for repeated trials. This suggested that task switching affected stage processes such as stimulus identification and response selection that occur before the onset of LRP. A further analysis of P300 confined the phenomenon of task switching to the processes that occur after the stimulus identification stage. Moreover, the results of additivity between task cueing and task switching suggested that the two factors affected a distinct stage of processing. A further implication would be that there was no switch-specific control process supporting the view of carry-over effect of switch cost.