108
SNR in the coupled network is double that of the uncoupled network. Figure 5.12 B shows
the SNR for an intrinsic noise source of unit variance in each rod over all practical com-
binations of spatial frequencies. Frequencies higher than the Nyquist sampling frequency
for individual rods are omitted. Figure 5.12 C shows a 2D projection of B, demonstrating
that for most frequencies able to be realized in the network, the coupled network shows an
increase in SNR.
5.7 Discussion
While it is known that rods are coupled in the salamander retina, the nature of the coupling
has been unclear due to conflicting reports of the coupling resistance. In this work we have
used two techniques, light stimuli, and electrical stimuli, to estimate the coupling resistance
between rods in the salamander retina.
5.7.1 Measurement of coupling with light
Other studies of electrical coupling between cones in the turtle retina used a similar tech-
nique, with a manually driven bar [69]. As mentioned by Detwiler and Hodgkin [32], a
problem with using a light stimulus to probe coupling is that the light stimulus is scattered
I
both by retinal tissue and the optics of the experimental setup. This scattering could lead
to an overestimate of the length constant A1 o, as it makes the stimulus appear wider to the
photoreceptors than it actually is. To circumvent this confounding factor, we measured the
profile of the scattered light in the retinal preparation using a cone photoreceptor as a detec-