102
rod network, its message is not easy to interpret. Spectral analysis via the Fourier transform
of these steady-state responses, on the other hand, reveals rod-rod couplings effect on im-
age frequency components. Because the frequency components of images may change over
the area covered by a single rod, the width of each rod and spacing between rods is also
taken into account. Rods are assumed to be spaced 16 microns apart with a diameter of 10
microns.
Figure 5.10 A shows the contributions of a single rod to the response of the entire net-
work with a 1 pA step test current. The response peaks at 0.141 mV (¾∏2D = 141 MΩ)
at the center rod, and coupling between rods allows signal spread to adjacent rods. Fig-
ure 5.10 B shows the response when the same rods are uncoupled. The response is 0.3 mV
(Rm = 300 MΩ). Because any arbitrary image can be decomposed into its frequency com-
ponents, examining the frequency response of both the coupled and uncoupled network
can demonstrate their behavior. Figure 5.10 C and D show example 2D and ID sinusoidal
gratings with frequencies of 50 microns per cycle.
Figure 5.11 shows the voltage response of a rod to varying frequencies in the x and y
dimension from the 2D Fourier transform of the patterns in figure 5.10. The amplitude of
the sinusoidal stimulus on the retina is a single rod’s response to a given frequency when
the peak over the rod’s center. Figure 5.11 A shows the response in the coupled network,
from 5.10 A, and 5.11 B shows the response in an uncoupled network. The diagonal repre-
sents stimuli of increasing frequency with x and y frequencies equal, as shown in figure 5.10
C. The uncoupled network has a frequency response (figure 5.11 B) with radial symmetry