77
Figure 2. Regions of interest (ROIs). Regions of interest were defined individually in each subject.
Colored voxels responded significantly more to somatosensory stimulation on the hands and feet
than to fixation baseline (p < IO*6). Different colors indicate different ROIs. Sl (green), primary
somatosensory cortex; Slfoot (blue), foot sub-region of SI; S2 (yellow), secondary somatosensory
cortex and associated areas; MST∕STP (orange), areas MST and STP.
Results
IneachsubjectzSoiDatosensorystiiDulationactivatedsoiDatosensorYcortexin
the postcentral gyrus a∏d parietal operculum and regions of visual association cortex in
lateral occipitotemporal lobe (Fig. ID). The response in these areas was used to train a
classifier, which in turn was used to decode the body site of somatosensory stimulation
for individual trials not used for training. The classifier prediction across all trials in a
single scan series is shown in Fig. IF. The classifier successfully predicted the correct
body site for stimulation for 85% of the trials (shown in green) and incorrectly classified
15% of the trials (shown in red)- Because classification was performed on each trial
separately, this level of prediction accuracy was highly unlikely to be due to chance. For
IOO hand and foot trials in the example scan series, the chance likelihood under the
binomial distribution of at least 85 correct trials was p < IO'38 (success probability per