Adaptation to Conflict Frequency: Non-Conflict Learning Is Not the Whole Story

In the Stroop task, smaller congruency effects (i.e., the color-naming difference between incongruent items, e.g., the word RED in the color blue, and congruent items, e.g., RED in red) are found in conditions in which incongruent items are frequent vs. infrequent. Although the traditional explanation for these “Proportion-Congruent effects” is that attention to task-relevant information is more focused in frequently-conflicting conditions (a process involving adaptation to conflict frequency), Proportion-Congruent paradigms typically have not controlled for the impact of more general learning processes, particularly 1) learning of word-response contingencies (contingency learning), 2) learning about the predictive nature of the stimuli (stimulus informativeness), and 3) learning about response rhythm in the task (temporal learning), processes which could produce the Proportion-Congruent effects obtained in most situations. The present research examined the possibility that those non-conflict learning processes are indeed the whole story in Proportion-Congruent effects. Several different approaches were used. First, the proportion of congruent and incongruent items in a list was manipulated in a variant of the Stroop task in which no individual stimulus was repeated, creating a situation in which neither contingency learning nor stimulus informativeness could have influenced task performance. Second, manipulating the proportion of neutral (i.e., consonant strings) and incongruent items in a list allowed the creation of a parallel situation in the classic color-word Stroop task. A Proportion-Congruent effect and a similar, Proportion-Neutral effect, emerged in both tasks even though contingency learning and stimulus informativeness could have played no role in producing those effects. Further, attempts to examine the influence of temporal learning failed to show any evidence of that process contributing to those effects either. The final set of experiments involved a congruency-proportion manipulation specific to individual words within the same list. Contrary to the idea that the Proportion-Congruent effect obtained in this situation results from contingency learning, a concurrent working memory load impaired contingency learning in a non-conflict color identification task but spared the Proportion-Congruent effect in the Stroop task, favoring a conflict-adaptation interpretation of this effect. Overall, these results support the existence of a process of adaptation to conflict frequency in the human control system