Figure 4.

In a simple real experiment, auditory N100(m) localization improves as data are combined across subjects, consistent with modeling predictions. (A) The activation maps (F-statistic values from sLORETA) for multiple subjects show activation in primary auditory areas, as well as point spread activation due to M/EEG inverse imaging. (B) The highest-activation (most statistically significant) points from each subject (colors) are more scattered than the highest-activation points from the average across subjects (black). The difference between the center of mass and an estimate of established N100(m) localization (C) and the spatial extent of the activation (D) is shown as a function of subjects. Mean ± 2 s.e.m. (across up to 2000 different random combinations of subjects) is plotted by the lines and shaded backgrounds. Note that these values were calculated the same way as the “centroid error” and “point spread” from simulations, but the lack of a ground truth activation location (or extent) made these labels inappropriate to assign to the ordinate axes.