Fig. 3.

Group-level effects in MEG (gradiometer) sensors during the training phase. (A) Main effect of sensory detail on sensor-space data. Topographic plots show statistical clusters showing a main effect of sensory detail (one plot for each cluster). Black circles within and the text directly above each scalp topography indicate the spatial and temporal extent of each cluster, respectively. The topographies themselves represent the difference in MEG response between 12-channel and three-channel speech, averaged over the temporal extent of the clusters. Waveforms represent the MEG response averaged over the spatial extent of a single selected cluster (indicated by the connecting black line). The width of waveforms represents ± two within-subject SEMs, similar to 95% confidence intervals. (B) The main effect of prior knowledge on sensor-space data. Topographies represent the difference in MEG response to speech presented after matching or mismatching written text, averaged over the temporal extent of significant clusters. (C) Right-hemisphere MEG cluster (at 312–792 ms) within which the neural effect of prior knowledge (matching/mismatching) correlated across subjects with the behavioral magnitude of perceptual learning of six-channel speech (speech recognition improvement post- vs. pretest). ch, channels; M, matching; MM, mismatching.