Figure 2.

Prediction suppression: Scrambled stimuli recruit greater activity in temporo-parietal networks than Sequential stimuli. (A) Overlay of Scrambled > Non-ordinal (blue), Sequential > Non-ordinal (green), and Scrambled > Sequential (red) contrasts shown in Figures 1B and C (p < 0.05 corrected). (B) Voxel counts of the clusters from the rIP and the rMTG from the previous two contrasts. In order to obtain a value subjectable to statistics, the contrasts were performed 30 times, each time using 70% of subjects (25 out of 35) (a bootstrapped voxel count). The resulting comparison shows that Scrambled stimuli recruit greater volumes than Sequential stimuli in the MTG and rIP (^***p < 0.001, repeated measures t-test). (C) Beta weights in the rIP are shown here averaged across the superior-inferior axis (z-axis) for all three conditions (for visualization only). The mask includes voxels that were found from either the contrast of Scrambled trials over non-ordinal trials, the Sequential over non-ordinal trials, or Scrambled over Sequential trials. Both amplitude and spatial extent of the rIP cluster decrease when ordinal stimuli are presented in a predictable order, as compared to a scrambled order.