Skip to main content
. Author manuscript; available in PMC: 2018 Oct 15.
Published in final edited form as: Neuroimage. 2017 Jun 20;180(Pt A):188–202. doi: 10.1016/j.neuroimage.2017.06.035

Figure 6.

Figure 6

Contributions of the DNN layers to dorsal and ventral stream predictions. Each column shows the distribution of the DNN layer contributions to the prediction accuracy for a single ROI. Colored bars within each column indicate the contribution to the prediction accuracy averaged over all voxels in that ROI. The contribution of the lowest (conv1) and highest (fc8) layers exhibit a clear counter-gradient organization. Contributions of intermediate DNN layers are also graded, but are much more uniformly distributed across ROIs.