Figure 4.

Match vs. mismatch trials for Experiments 1, 3, and 4 (pooled over frequencies). Responses following correct predictions (match) in Experiment 1 are 117 ms faster on average compared to incorrect predictions (mismatch) in the first two blocks; after the introduction of the selective response deadline for the medium frequency stimulus this difference is reduced to 60 ms on average. In Experiment 3, using invalid cues (similar to the predictions of Experiment 1), the difference between match and mismatch trials averages 34 ms at the beginning and is down to 17 ms with the additional task demand. Experiment 4 shows no reduction in this mismatch effect, with 98 ms before and 100 ms after the introduction of the deadline on average.