Figure 11:

Effects of various manipulations on the extent to which TE representations differentiate from V1. For all plots, Intact is the same result shown in Figure 6 from the intact model for ease of comparison (panel a is missing V3 and DP dorsal pathway layers). All of the following manipulations significantly impair the development of abstract TE categorical representations (i.e., TE is more similar to V1 and the other layers). a) Dorsal Where pathway lesions, including lateral inferior parietal sulcus (LIP), V3, and dorsal prelunate (DP). This pathway is essential for regressing out location-based prediction errors, so that the residual errors concentrate feature-encoding errors that train the What pathway. b) Allowing the deep layers full access to current-time information, thus effectively eliminating the prediction demand and turning the network into an auto-encoder, which significantly impairs representation development, and supports the importance of the challenge of predictive learning for developing deeper, more abstract representations. c) Reducing the strength of Hebbian learning by 20% (from 2.5 to 2), demonstrating the essential role played by this form of learning on shaping categorical representations. Eliminating Hebbian learning entirely (not shown) prevented the model from learning anything at all, as it also plays a critical regularization and shaping role on learning.