Cross-task congruency sequence effect without the contribution of multiple expectancy.

The confound-minimized cross-task design has been widely used to examine the characteristics of top-down cognitive control underlying the congruency sequence effect (CSE) without feature integration and contingency learning confounds. The present study reanalyzed our previous data obtained with the confound-minimized cross-task design, this time including the preceding congruency repetition type, to examine whether the cross-task CSE is confounded by feature integration from two-back (n-2) trials or multiple expectancies regarding the congruency and the congruency repetition type of the upcoming trial. As a result, the cross-task CSE interacted with the arbitrariness of S-R mapping or response mode regardless of the preceding congruency repetition type, indicating the contribution of top-down control triggered by conflict. Feature integration from n-2 trials, but not multiple expectancies, was found to have a lingering effect on the sequential modulation of the congruency effect between previous and current trials. However, because the influence of feature integration operated in opposite directions depending on the preceding congruency repetition type, the contribution of feature integration to the cross-task CSE can be minimized when the combined datasets of trials following a congruency repetition trial and those following a congruency alternation trial are analyzed. These findings are consistent with recent perspectives on cognitive control, which posit that top-down cognitive control and bottom-up feature integration operate independently to optimize task performance.