Figure 4.

Different degrees of angioscotoma representation similar to those observed experimentally (Fig. 1) are produced in two-dimensional cortices using the elastic net algorithm. A–L, Different degrees of representation were generated by changing the ratio l/K_b, which controls how segregated the OD columns are at birth. A–L, Two-dimensional versions of the simulations in Figure 3. K, M–O, Representations of different-shaped angioscotomas in cortex with l = 0.05 and K_b = 0.038. These show that the degree of angioscotoma representation is independent of the specific shape of the angioscotoma. K, P–R, Angioscotoma representations in different cortices with l = 0.05 and K_b = 0.038 produced in the elastic net by starting the net from different random initial configurations. These show that the influence of the angioscotoma is largely independent of the particular pattern of columns that would have formed otherwise. All other parameters were the same in each case. For all simulations, annealing was continued at the same rate for 10 iterations after angioscotoma introduction. The red crosses indicate the cortical locations of the feature points that are deprived after birth. Black represents afferents from one retina, and white represents afferents from the other.