Relationship between Psychophysical and fMRI Learning Effects

<p>fMRI data and the corresponding psychophysical response for low-salience (A) and high-salience (B) shapes. For each individual subject, we plotted a behavioral learning index (percent correct for trained minus percent correct for untrained stimuli) and an fMRI learning index (percent signal change for trained minus percent signal change for untrained stimuli) after training. Positive values indicate stronger responses for trained than untrained shapes, whereas negative values indicate lower responses for trained than untrained shapes. For low-salience shapes the regression analysis was significant in early visual areas (V1: <i>r</i> = 0.62, F_1,21 = 13.65, <i>p</i> = 0.001; V2: <i>r</i> = 0.42, F_1,21 = 6.65, <i>p</i> < 0.05; Vp: <i>r</i> = 0.50, F_1,21 = 7.62, <i>p</i> = 0.01; V4: <i>r</i> = 0.50, F_1,21 = 8.91, <i>p</i> < 0.01) and the LOC subregions (LO: <i>r</i> = 0.57, F_1,21 = 9.84, <i>p</i> < 0.01; pFs: <i>r</i> = 0.51, F_1,21 = 7.18, <i>p</i> = 0.01). For high-salience shapes the regression was significant only in the LOC subregions (LO: <i>r</i> = 0.56, F_1,15 = 6.69, <i>p</i> < 0.05; pFs: <i>r</i> = 0.61, F_1,15 = 8.58, <i>p</i> = 0.01) but not in the early visual areas (V1: <i>r</i> = 0.17, F_1,15 < 1, <i>p</i> = 0.51; V2: <i>r</i> = 0.25, F_1,15 < 1, <i>p</i> = 0.34; Vp: <i>r</i> = 0.28, F_1,15 = 1.27, <i>p</i> = 0.27; V4: <i>r</i> = 0.15, F_1,15 < 1, <i>p</i> = 0.56).