Our performance on a task decreases when the task is in a dual-task situation than when it is in isolation. An important experimental setting for dual-task situation is the psychological refractory period (PRP) paradigm, and the dual-task performance decrements in the PRP paradigm are referred to as PRP interference. The standard response-selection bottleneck (RSB) models state that the response-selection stage of the second task (T2) cannot start until the response-selection stage of the first task (T1) finishes, resulting in the PRP interference. Contrary to the prediction of RSB models, several researchers have found T2's modulations on T1's performance, and have suggested that T1's selection-related processes are affected by T2's selection-related processes, referred to as backward crosstalk effects. The locus of backward crosstalk effects is not clear, however, because RTs were measured in most previous studies. By using semantically unrelated stimuli and responses and by measuring T1's lateralized readiness potential, we examined the locus of backward crosstalk effects. We found that the interval between T1's stimulus onset and the stimulus-locked LRP onset was affected, suggesting T2's response selection starts before T1's selection is complete. The present result provided electrophysiological evidence focusing on T1's changes in favor of the hypothesis of parallel response selection in the PRP paradigm.