Random Wiring, Ganglion Cell Mosaics, and the Functional Architecture of the Visual Cortex
Fig 9
Iso-orientation domain layouts obtained from PIPP RGCs with the statistical connectivity model.
A Generating orientation domain layouts from PIPP RGCs in the statistical connectivity model [83]. Top: Inset of a PIPP RGC mosaic (see Methods). Black (white) dots represent OFF (ON) cells. Middle top: unfiltered layout with RGC mosaics overlaid. Middle bottom: thresholded layout with RGC mosaics overlaid. Bottom: thresholded and smoothed layout (β = 0) with RGC mosaics overlaid. Scale bar indicates retinal distances, assuming PIPP parameters as in [49]. B Left: larger region of the unfiltered layout shown in A (black square). Scale bar indicates retinal distances. Right: normalized amplitude spectrum of unfiltered layout shown on the left. C Thresholded and smoothed layout (top) and corresponding amplitude spectrum (bottom) for filter function parameters (see Eq (15)) β = 0 (left), β = 2 (middle), and β = 10 (right). Scale bar indicates cortical distances, assuming cortical magnification factor ≈ 1, and Λ = 0.9mm (see Eq (17) and text). Red circles indicate kc = 2π/Λ. Black square indicates inset in A, white square indicates Λ2. D As C but filtered with Fermi band pass filters [23]. White square (top) indicates Λ = 0.68mm, the column spacing as measured by wavelet analysis. Red circles (bottom) indicate low pass (1.2 mm) and high pass (0.3 mm) position. Pinwheel densities are stated with standard error of the mean. E Left: Analytically predicted pinwheel density of orientation domain layouts derived from Gaussian random fields [51] as a function of filter parameter β and spatial scale (see text). Right: Pinwheel density of orientation domain layouts obtained from PIPP mosaics with the statistical connectivity model as a function of filter parameter β and spatial scale. Numbers 1–3 indicate parameter choices displayed in C.