We investigated the Gratton effect and the proportion congruency effect using a parity/magnitude task-switching paradigm. In our study, participants decided the parity or magnitude of a digit. These tasks alternated across trials. Congruency was defined as the match between the stimulus-response rules of the current task with the other task. Therefore, there was no irrelevant dimension, no conflict within the stimulus, or no focus on the relevant dimensions. Conflict-driven control directs the conflict that arises when multiple, competing rules are held in working memory. Importantly, the stimulus-response contingency remains the same across different levels of conflict. We observed both the proportion congruency effect and the Gratton effect in both reaction times and error rates. Our results suggest that the conflict-driven control incorporates conflict between rules, arising from holding conflicting rules in immediate memory. Contingency learning alone cannot fully explain the the proportion congruency effects observed in our task-switching paradigm.