Figure 1.

The network architecture of a large scale network model. Small grey circular patches in (a) represent 15 by 15 local components in the network. These local components are placed horizontally on a plane parallel to cortical surface which mimic hypercolumns in V1. The spatial range of activation zone, or the number of activated E-I units in the model, is dependent on the size of external visual stimulus. Each local component in (b) consists of excitatory (E) and inhibitory (I) component connected with local connection W_EE^RC, W_EI^RC, W_IE^RC, and W_II^RC. In the cortical space, the E and I components receive feedforward input (W_EL, W_IL) from subcortical regions (LGN) that are driven by visual stimuli, horizontal inputs (W_EE^HC and W_IE^HC) from other E-I units and receives feedback (W_EG, W_IG) from higher visual cortex (G). The higher visual cortex (G) receives feedforward (W_GE) input from V1 as well.