The kinetics, gain, and reliability of light responses of rod and cone photoreceptors are important determinants of overall visual sensitivity. In voltage recordings from photoreceptors in an intact primate retina, rods were found to be functionally isolated from each other, unlike the tightly coupled rods of cold-blooded vertebrates. Cones were observed to receive excitatory input from rods, which indicates that the cone pathway also processes rod signals. This input might be expected to degrade the spatial resolution of mesopic vision.

Whole-cell voltage and current recordings were obtained from red and green cone photoreceptors in isolated retina from macaque monkey. It was demonstrated previously that the cone photovoltage is generated from two sources, phototransduction current in the cone outer segment and photocurrent from neighboring rods. Rod signals are likely transmitted to cones across the gap junctions between rods and cones. In this study, the "pure" cone and rod components of the response were extracted with rod-adapting backgrounds or by subtracting the responses to flashes of different wavelength equated in their excitation of either rods or cones. For dim flashes, the pure cone component was similar in waveform to the cone outer segment current, and the rod component was similar to the photovoltage measured directly in rods. With bright flashes, the high frequencies of the rod signal were filtered out by the rod/cone network. The two components of the cone photovoltage adapted separately to background illumination. The amplitude of the rod component was halved by backgrounds eliciting ϳ100 photoisomerizations sec Ϫ1 per rod; the cone component was halved by backgrounds of 8700 photoisomerizations sec Ϫ1 per cone. Coupling between rods and cones was not modulated by either dim backgrounds or dopamine. Voltage noise in dark-adapted cones was dominated by elementary events other than photopigment isomerizations. The dark noise was equivalent in magnitude to a steady light eliciting ϳ3800 photoisomerizations sec Ϫ1 per cone, a value significantly higher than the psychophysical estimates of cone "dark light." Key words: rod; cone; photoreceptor; primate; adaptation; couplingCertain aspects of human vision, such as color sensitivity, are determined by the process of light absorption and phototransduction in cone outer segments Schnapf et al., 1987). But other aspects of vision, such as temporal sensitivity and light adaptation, cannot be explained by properties of the outer segment and must therefore be determined principally by processes downstream (Schnapf et al., 1990). The first downstream locus of signal processing in the retina is the photoreceptor inner segment. Here the photocurrent is shaped by voltageactivated conductances (Barnes, 1994) and combined with synaptic inputs from neighboring neurons (Baylor et al., 1971). The purpose of the present work was to understand how the cone inner segment modifies the photocurrent and how these modifications affect vision. To address these questions, we recorded photovoltage from cone inner segments in freshly isolated retina from macaque monkeys.Psychophysicists describe two distinct retinal mechanisms by which rod signals are processed in parallel. One mechanism is slow and sensitive to dim light, and the second is faster and operates at mesopic light levels (Conner and MacLeod, 1977;Conner, 1982). It has been suggested that the rod-to-bipolar synapse connects rods to the more sensitive mechanism and that gap junctional coupling between rods and cones connects rods to the second mech...

SUMMARY1. Photocurrents were recorded with suction electrodes from rod photoreceptors of seven humans.2. Brief flashes of light evoked transient outward currents of up to 20 pA. With increasing light intensity the peak response amplitude increased along an exponential saturation function. A half-saturating peak response was evoked by approximately sixty-five photoisomerizations.3. Responses to brief dim flashes rose to a peak in about 200 ms. The waveform was roughly like the impulse response of a series of four to five low-pass filters.4. The rising phases of the responses to flashes of increasing strength were found to fit with a biochemical model of phototransduction with an 'effective delay time' and 'characteristic time' of about 2 and 800 ms, respectively.5. Spectral sensitivities were obtained over a wavelength range from 380 to 760 nm. The action spectrum, which peaked at 495 nm, followed the template described for photoreceptors in the macaque retina. Variation between rods in the position of the spectrum on the wavelength axis was small.6. The scotopic luminosity function derived from human psychophysical experiments was found to agree well with the measured rod action spectrum after adjustments were made for lens absorption and photopigment self-screening in the intact eye.7. Responses to steps of light rose monotonically to a maintained level, showing little or no relaxation. Nevertheless, the relationship between light intensity and steady-state response amplitude was shallower than that expected from simple response saturation. This is consistent with an adaptation mechanism acting on a rapid time scale.8. Flash sensitivity fell with increasing intensities of background light according to Weber's law. Sensitivity was reduced twofold by lights evoking about 120 photoisomerizations per second. Background lights decreased the time to peak and the integration time of the flash response by up to 20%.

Membrane voltage was recorded in rod photoreceptors in retina isolated from macaque monkey. The size of the single photon response and the magnitude of membrane voltage fluctuations were assessed in dark- and light-adapted retina. The "dark light" rate I(D), defined as the rate of spontaneous photopigment isomerizations that would produce a variance equivalent to that of the noise measured in the dark, was calculated after matched filtering. The average value of 0.08 s(-1) fell at the higher end of psychophysical estimates of dark light in human observers. In light-adapted rods the photon response decreased in amplitude and duration, and the magnitude of the voltage fluctuations increased with increasing background light intensity. The signal-to-noise ratio (SNR) for single rods was defined as the ratio of the peak amplitude of the photon response to the standard deviation of the noise fluctuations. The signal-to-noise ratio for dark-adapted rods SNR(D) was about 7. With increasing background intensity I, the SNR fell as SNR(D)(1 + I/I(D))(-1/2). This function may account for the increment thresholds measured with small brief test flashes in human psychophysical experiments.

Microelectrodes were used to record from red and green cones of the turtle Pseudemys scripta elegans. The purpose of this study was to determine the action spectra of the red and green cone photopigments, and to look closely for direct interactions between the two cone classes. An isolated retina preparation was employed so that cones could be stimulated from the outer segment side, thereby avoiding the oil droplets that reside in the inner segments of many cones and normally filter incident light. In agreement with some previous electrophysiological studies, we found little evidence for significant direct connections between red and green cones. Exceptions to this rule are noted and discussed. Measurements indicate that this result does not appear to be due to a general loss of cone connectivity in the isolated retina preparation. Action spectra of the cone photopigments differed markedly from action spectra reported for cones in the eyecup preparation. In contrast to cones in the eyecup, cones in the isolated retina showed higher short-wavelength sensitivity and had action spectra that were adequately described by photopigment nomograms. A model of cone optical properties suggests that in the eyecup up to about 40% of the light that reaches a cone outer segment may do so without first passing through an oil droplet.