Figure 1.

Task used in Experiment 1. On each trial, participants had to verbalize their prediction for the upcoming stimulus (in this case “star”). After 2500 ms the stimulus appeared on screen (in this case a house, signifying a mismatch) and participants had manually respond by pressing one of three keys. The next trial started 500 ms after the response. For any given participant, only one of the two stimulus features (shape, color) was relevant throughout the task (in this case, both predictions and responses pertained to the shape of a stimulus).