Abstract
Behaviour that is assumed to be guided by strategy can, in fact, be based on the implicit learning of regularities in the environment. We demonstrate this point in the context of a Stroop experiment. It has been shown previously that performance on this measure of cognitive control varies as a function of the relative proportions of congruent and incongruent trials in a block. Here we provide evidence that this modulation of performance is largely based on implicit, rather than explicit, knowledge of these proportions. This result has important implications for our understanding of cognitive control.
Notes
1 A total of 5 participants completed 189 sessions; 1 missed a 25% block, 2 missed a 50% block, and 2 missed a 95% block.
2 An analysis at the level of individual trials yields the same statistical pattern of results.
3 One participant is excluded from this analysis. He incorrectly answered Question 1 for most of the blocks, despite high certainty on Question 2 and accurate estimates on Question 3. Recoding his responses and including him does not affect the results.
4 This analysis ignores the repeated measures component because each participant has data at only 4–23 blocks (mean = 11, SD = 7). In addition, the range of proportion values represented varies widely across participants from 15–85 (mean = 51, SD = 26). That said, the repeated measures analysis finds a congruency effect (.158 ≤ β≤ .384) and a congruency by proportion interaction in the right direction (–.048 ≤ β≤ .224), p = .174. The participants who fail to show an interaction in this analysis are those with the smallest range.