Multiple expectancies underlie the congruency sequence effect in confound-minimized tasks.

The congruency sequence effect (CSE) occurs when the congruency effect observed in tasks such as the Eriksen flanker task is smaller on trials preceded by an incongruent trial relative to trials preceded by a congruent trial. The CSE has been attributed to a range of factors including repetition expectancy, conflict monitoring, feature integration, and contingency learning. To clarify the debate surrounding the CSE and the mechanisms underlying its occurrence, researchers have developed confound-minimized congruency tasks designed to control for feature-integration and contingency-learning effects. A CSE is often observed in confound-minimized tasks, indicating that the effect is driven by repetition expectancy, conflict monitoring, or a combination of the two. Here, we propose and test a variant of the repetition expectancy account that emphasizes how multiple expectations can be formed simultaneously based upon the congruency type (congruent vs. incongruent) and the congruency repetition type (congruency repetition vs. congruency alternation) of the most recent trial. Data from confound-minimized versions of the prime-probe task were found to support this novel account. Data from confound-minimized versions of the Eriksen flanker, Simon, and Stroop tasks indicate that feature-integration confounds often remain in these tasks, potentially undermining the conclusions of previous work. We discuss the implications of these findings for ongoing theoretical debates surrounding the CSE. • The CSE observed in confound-minimized tasks is modulated by the congruency repetition type of the previous trial • The multiple-expectancies account is introduced to explain the effect of the previous trial's congruency repetition type • Repetition-priming confounds remained in versions of the flanker, Simon, and Stroop tasks designed to minimize confounds • No evidence of lingering repetition-priming confounds were observed in confound-minimized versions of the prime-probe task [ABSTRACT FROM AUTHOR]