## Results: 7

1.

Rod tracer coupling. A,B: Combined confocal fluorescence and DIC images of the outer nuclear layer after single rod injections with Neurobiotin (green). Scale bar, 10 μm. A: Rod is not tracer coupled to neighbors. B: Four tracer-coupled rods. C: Histogram of tracer coupled pool sizes for 32 Neurobiotin filled rods.

2.

Guinea pig rod spectral sensitivity. Points plot the mean sensitivity of 9 rods for 430, 500, 570, and 660 nm wavelength stimuli. Standard deviations are smaller than symbol diameters. The curve is equation 6 from Baylor et al. (1987) after correction for photopigment self-screening, with λ

_{max}= 496 nm.3.

Rod-rod gap-junctional conductance. A: Changes in junctional current in a voltage clamped rod in response to voltage steps applied to a neighboring rod. Bandwidth, DC–100 Hz. B: Peak junctional current

*I*vs. junctional voltage_{J}*V*for the rod pair in A averaged from 25–50 ms after onset of the voltage step. The slope of the best fitting line gives a junctional conductance_{J}*G*of 572 pS. C: Histogram of_{J}*G*from 22 rod pairs. Measurements were corrected for series resistance. D: Schematic of indirect current path between two voltage-clamped rods, 1 and 2, via rod 3. E: Rod 4 provides an additional indirect path. F: In a hexagonal network, a maximum of two indirect paths with single intervening rods are possible._{J}4.

Coupling reduces the variability of dim flash responses. A,B: Rod voltage responses recorded in two rods to a series of dim flashes evoking ~1 R

^{*}. Tick marks indicate flash timing. Bandwidth, DC–5 Hz. C,E: Response mean (C) and variance (E) of rod in A to 50–100 flashes evoking 1.5, 2.8, and 5.5 R^{*}. D,F: Peak amplitude of the mean (μ) and variance (σ^{2}) as a function of R^{*}, measured from rod A (○) and B (▽). The lines near the data points are the best-fit linear functions passing through the origin. From the slopes of the lines,*N*was calculated as 0.9 (○) and 10.2 (▽). G: Histogram of*N*from 14 rods.5.

Primate rod network model. A: Average change in membrane current in a macaque rod in response to a pulse in holding potential from −50 mV to −51 mV. The response of an isolated rod is expected to decay exponentially; from the best fitting exponential function (smooth curve) the calculated model parameters were

*R*= 1.67 GΩ,_{m}*R*= 1.15 GΩ, and_{L}*L*= 101 MH. Average of 100 responses. Timing of the 300 ms voltage pulse indicated by the bar below the current trace. Capacitive transients were reduced by filtering; bandwidth, DC-100 Hz. B: Schematic of the two-rod network with perfect coupling (Hornstein et al., 2005). Voltage transfer ratios relative to rod 1 are*w*_{1|1}=*w*_{1|2}= 0.5. C: Schematic of the four-rod network used in psychophysical modeling; voltage transfer ratios relative to rod 1 are*w*_{1|1}= 0.624,*w*_{1|2}= 0.154,*w*_{1|3}= 0.154,*w*_{1|4}= 0.068.6.

Coupling effects on human visual detection. A: Dark-adapted detection threshold for coupled rod network

*T*relative to uncoupled rod network_{C}*T*, as a function of stimulus diameter on the retina. Points plot model calculations with β=2.5. For small pupil sizes, the stimuli are all optically realizable, but for the larger pupils of dark-adapted eyes, stimuli <~0.06 deg diameter are significantly broadened by the optical point spread function of the eye. Arrows indicate the threshold ratios calculated for a point source stimulus after accounting for optical blur of an eye with a 6 mm or 8 mm pupil. B: Signal-to-noise ratio of coupled rod network SNR_{U}_{C}relative to uncoupled rod network SNR_{U}as a function of background light intensity, with β=2.5. Illumination in B is uniform across the detector pool. SNRs were modeled based on the psychophysical threshold data of a rod monochromat given in Sharpe et al. (1992). The curves in A and B are empirical functions fitted to the data points.7.

Hexagonal network modeling. A: DIC photomicrograph of the inner segment layer illustrating the roughly hexagonal packing of the guinea pig rod mosaic. Scale bar, 10 μm. B: Schematic of a resistive rod network model: a field of hexagonally packed rods grouped into several discrete coupled networks with varying connectivities.

*R*is rod membrane resistance;_{m}*R*is gap junctional resistance. C: Network model with complex impedance._{j}*C*is membrane capacitance; electrical equivalent of voltage-dependent conductances represented by an inductance_{m}*L*and shunt resistance*R*. D: Average change in membrane potential in a guinea pig rod in response to a 1 pA current injection. Smooth curve gives the expected waveform of the voltage response of an isolated rod (equation 2 in Baylor et al., 1984a); best fitting model parameters were_{L}*R*= 1.65 GΩ,_{m}*R*= 3.50 GΩ,_{L}*L*= 450 MH, and*C*= 5 pF. Average of 100 responses._{m}*V*= −35 mV. Timing of the 300 ms current pulse indicated by the bar below the voltage trace. Bandwidth, DC-500 Hz._{rest}