The HCP language task (story vs baseline) beta maps and their spatial gradients. Beta maps (rows 1 and 3) and gradient maps (rows 2 and 4) are from two independent groups of 210 HCP subjects, "210P" (rows 1 and 2) and "210V" (rows 3 and 4). Because of the large number of high quality HCP subjects, the beta maps are very similar across the two groups, and the strong gradients in the beta maps are also very similar. Also shown are white contours of a Bonferroni corrected significance threshold across all 91282 grayordinates (z+/- ~5). Two things are apparent: 1) Because of the large amount of high quality data, most of the brain is either significantly activated or deactivated. Thus the statistical threshold is not particularly biologically meaningful (a point about statistical thresholds that generally applies). At the same time, the statistical threshold is also not strongly reproducible, in spite of the large amount of high quality data (highlighted ellipses show large differences in the area of activation classified as "significant" that are not particularly impressive when viewing the unthresholded beta maps). In contrast, the strong gradients in the beta maps are much more reproducible, are likely also more biologically meaningful, and hence provide a better substrate for defining regions of activation or comparing across studies.
Surface Mesh:32k fs LR, Registration:MSMAll, Species:Human, Modality:Myelin Map, Modality:T2-weighted, Modality:T1-weighted