Actually, when I applied the linear interploration using "@autl_tlrc -rmode linear" the differences between adwarp became MORE pronounced (see below). This makes me wonder whether it was really the interpolation mode that made the difference between @auto_tlrc and adwarp...
-- At sub-brick #2 'Mean' datum type is float: -13.7136 to 13.833
keywords = negenh_auc_perc+orig[0]
-- At sub-brick #3 'Mean' datum type is float: -7.54116 to 14.4384
keywords = negmain_auc_perc+orig[0]
-- At sub-brick #4 'Mean' datum type is float: -9.77812 to 14.8726
keywords = negsup_auc_perc+orig[0]
-- At sub-brick #5 'Mean' datum type is float: -8.51547 to 19.3278
keywords = neumain_auc_perc+orig[0]
-- At sub-brick #6 'Mean' datum type is float: -7.77349 to 7.33811
keywords = negenh-negmain_auc_perc+orig[0]
-- At sub-brick #7 'Mean' datum type is float: -3.17449 to 6.49137
keywords = negmain-negsup_auc_perc+orig[0]
-- At sub-brick #8 'Mean' datum type is float: -8.33944 to 11.4842
keywords = negenh-negsup_auc_perc+orig[0]
-- At sub-brick #9 'Mean' datum type is float: -6.42679 to 5.54711
keywords = negmain-neumain_auc_perc+orig[0]
-- At sub-brick #10 'Mean' datum type is float: -26.6289 to 39.0937
keywords = neg_mean_auc_perc+orig[0]
-- At sub-brick #11 'Mean' datum type is float: -25.5145 to 24.6227
keywords = neg-neu_auc_perc+orig[0]