The influence of word reading on Stroop color naming decreases as a function of the proportion of test items that are incongruent. This proportion-congruent effect is usually ascribed to strategies (e.g., maintaining task set) that operate at a general level to moderate the extent to which participants are influenced by word reading. However, in three experiments, effects at the level of specific items were found. Interference and facilitation were smaller for color names usually presented in an incongruent color than for color names usually presented in their congruent colors. This item-specific proportion-congruent manipulation affected the process dissociation (PD) estimate of the influence of word-reading processes but not that of color-naming processes. The results (1) indicate that item-specific, as opposed to general, mechanisms can reduce the influence of word-reading processes on Stroop performance and (2) demonstrate the PD procedure's utility in studying Stroop phenomena
