Figure 3. Effects of learning stage and agency on acquisition sound ERPs and topographical cluster plots. a.Grand-averaged ERP wave forms of the late learning stage (blue ) and the early learning stage (red ) acquisition sounds, plus the difference wave (early subtracted from late; black ) at the Fz electrode. The P3a time window is highlighted. The asterisk indicates a significant difference across learning stages. Below, a series of 5 topographical plots shows the difference between late and early learning stage (late - early) across time and sensor space. Electrodes belonging to the significant negative cluster are highlighted in white.b. Grand-averaged ERP wave forms of the agent condition (blue ) and the observer condition (red ) acquisition sounds, plus the difference wave (observer subtracted from agent;black ) at the Pz electrode. The P3b time window is highlighted. The asterisk indicates a significant difference across agency conditions. Below, a series of 4 topographical plots shows the difference between agent and observer condition (agent - observer) across time and sensor space. Electrodes belonging to the significant positive cluster are highlighted in black.

Test sounds

Congruency between movement and sound triggered a late positive deflection that was absent in incongruent sounds (figure 4). The cluster-based permutation test comparing congruent and incongruent test sounds detected a significant positive cluster (T = 4946.6, p = .004; 240 ms to 400 ms) that starts at occipito-parietal electrodes and spreads across the entire skull. The cluster temporally encompasses the P2 and P3 components, revealing overall more positive amplitudes for congruent sounds. The targeted-component ANOVAs found a significant main effect of congruency on the P3 component in Fz [F(1,22) = 12.342, p=.001, ηp2 = 0.36] and Pz [F = 20.589, p < .001, ηp2 = 0.48], corresponding to the P3a and the P3b. The effects of congruency were not modulated by agency or by learning stage.
We found no effects of agency nor learning stage on test sounds, and no interactions. Both the cluster-based approach and the targeted-component analysis yielded no significant results.