Fig_2.tif (503.21 kB)

Four example items.

Download (0 kB)
posted on 10.06.2015, 03:25 by Aya Hatano, Taiji Ueno, Shinji Kitagami, Jun Kawaguchi

The 3rdthrough 5thcolumns show the representations (Hinton diagrams) in each layer. The leftmost column shows the item ID numbers for each face pattern. The second column shows the unique combination of the specific (shown in bold) features of each face, which the model could not discriminate by the activation patterns of the verbal layer (i.e., 5th column). The third column shows the retinotopic patterns of each item. The fourth column shows the target patterns of the visual image layer. The fifth and sixth columns show the activation patterns (and their labels) in the verbal layer.