Figure 2.

General model architecture. (A) The inhibitory cortical population corresponds to a regular lattice of units in cortical space. (B) As in A but for the excitatory population. (C) Thalamic neurons are modeled as simple difference-of-Gaussian filters followed by a threshold-linear transfer function. (D) The intra-cortical connectivity between the excitatory and inhibitory populations and their transmission delays. The intra-cortical connectivity is the only aspect of the general architecture presented here that changes between the different model variants explored in this study and each variant is detailed in Figure 3. (E) The RF centers of the LGN neurons form a regular latice across the visual space covered by the model. This retinotopic mapping of connection fields between layers of the model is also maintained in the thalamo-cortical projection (lines between D,E). These thalamo-cortical projections are the only connections in the model that undergo Hebbian adaptation during simulated visually driven development.