86
accurate than S2 (61% vs. 50%, p = 0.01). Sl and S2 both contain multiple subdivisions
(Eickhoff, et al. 2006a; Kaas, et al. 1979) and in future experiments at higher resolution
it will be important to study the ability of these subdivisions to discriminate
Somatosensorystimuli.
Previous studies have reported responses to somatosensory stimuli in "visual"
cortex (Amedi, et al. 2001; Sathian, et al. 1997). In particular, tactile responses have
been reported in a region of lateral occipital-temporal cortex that contains area MST
and the possible human homolog of area STP (Beauchamp 2005a; Beauchamp, et al.
2008; Beauchamp, et al. 2007; Blake, et al. 2004; Hagen, et al. 2002; Ricciardi, et al.
2007). Previous MVPA studies have shown that MST and nearby areas can decode the
direction of motion, but not the orientation, of visual stimuli (Kamitani and Tong 2005;
Kamitani and Tong 2006). Here, we extend these findings by showing that fMRI
■ ■ ■ ■ ∙ ■ ■ ■ / ■
activation patterns in MST∕STP are able to decode information about the hand of
somatosensory stimulation. Inactivation of monkey MST interferes with visually-guided
hand movements (Hg and Schumann 2007) and transcranial magnetic stimulation of
human MT∕MST reduces reaching accuracy (Whitney, et al. 2007). While visual signals
provide an accurate initial targeting signal during reaching movements, determining
whether a target has actually been touched is most easily accomplished by the
somatosensory system. Consistent with this idea, MVPA of area MST∕STP was able to
determine the location of hand stimulation (left vs. right) with performance far above
chance. However, MST∕STP was not able to decode the finger of touch for fingers on the
same hand, suggesting that tactile inputs into MST∕STP are not highly specific, perhaps