Fig. 3.

SMLR weight vector discrimination maps for the multi-class classifier discriminating between all drug conditions. Top panel: weight vector for MPH, middle panel: weight vector for ATX, bottom panel: overlapping voxels. For the top two panels, positive coefficients (red color scale) indicate a positive contribution to the prediction for each class and negative coefficients (blue color scale) indicate a negative contribution. For the bottom panel, red indicates voxels with non-zero coefficients in the MPH weight vector, blue indicates voxels with non-zero coefficients in the ATX weight vector and yellow indicates voxels with non-zero coefficients in both weight vectors. Note that the weight vector for PLC is fixed to zero and that the scale for weight vector coefficients is arbitrary. The right hand side of each image corresponds to the participants' right side and numerals in white text indicate Z-coordinates in Talairach space.