Table 7.
MKL modelling of the comparison of ‘faces’ vs ‘houses’, based on the AAL atlas
| Region | Contribution (%) | Size | ER |
|---|---|---|---|
| Fusiform_L | 21.342208 | 617 | 115.29 |
| Occipital_Mid_L | 20.902485 | 839 | 114.92 |
| Lingual_R | 20.323188 | 642 | 114.50 |
| Lingual_L | 13.591652 | 597 | 108.29 |
| Fusiform_R | 10.874692 | 687 | 102.96 |
| Cingulum_Post_L | 5.947777 | 143 | 110.83 |
| Frontal_Inf_Oper_R | 4.831543 | 373 | 101.17 |
| Parietal_Sup_L | 1.248777 | 401 | 81.54 |
| Occipital_Sup_L | 0.464520 | 314 | 45.17 |
| Caudate_L | 0.152138 | 285 | 4.58 |
| Occipital_Mid_R | 0.136176 | 441 | 9.00 |
| Hippocampus_R | 0.109758 | 284 | 13.54 |
| Cingulum_Ant_L | 0.074190 | 402 | 18.00 |
| Thalamus_L | 0.000894 | 290 | 4.50 |
Regions ranked further (i.e. rank >14) have a perfectly null contribution to the model. ER stands for Expected Ranking. The region size is displayed in voxels. The weights were averaged across folds (except for computing ER). A region label ending in ‘L’ (resp. ‘R’) means left (resp. right) hemisphere region