Content uploaded by Kristian Donner
Author content
All content in this area was uploaded by Kristian Donner
Content may be subject to copyright.
Visual latency and brightness: An interpretation based on the responses of rods and ganglion cells in the frog retina
Abstract
Rod and cone photoresponses in a variety of species have been accurately described with linear multistage filter models. In this study, the response latency and initial coding of intensity at two higher levels of visual processing are related to such photoreceptor responses. One level is the retinal output (spiking discharges from frog ganglion cells, based on experimental data reported here), the other is the perceptual level in humans (psychophysical latency and brightness functions, based on data from the literature). Photoreceptor responses are described with the "independent activation" model of Baylor et al. (1974). The intensity dependence of the early ganglion cell discharge, its latency and initial impulse frequency, is shown to follow from such a waveform, assuming that 1) latency L = l + D, where l is the time it takes for the rod response linearly summed over the ganglion cell's receptive field to reach a criterion amplitude, and D is a constant delay; and 2) the initial frequency (below saturation) is proportional to the steepness of rise of the summed rod response at time l. It is shown that the intensity dependences of 1) human visual latency and 2) brightness sensation, including effects of stimulus area and duration, are accounted for by the same model. The predicted functions are not power functions of intensity, but approximate such over wide ranges. Thus, a large body of psychophysical data is explained simply by the waveform of photoreceptor responses.
... Such experiments performed under a series of dim backgrounds, at different temperatures, and with varying worm velocities and sizes have provided several fundamental insights. Snapping precision is mechanistically limited by the light-intensity-dependent latency of the retinal GCs [203], but is supported by a predictive component evident when GC latencies become long compared with worm velocity [26]. GC latencies, in turn, are long at low illumination levels, because they are determined by the slow response kinetics of the rods, which on the other hand gives the advantage of extensive temporal integration. ...
... GC latencies, in turn, are long at low illumination levels, because they are determined by the slow response kinetics of the rods, which on the other hand gives the advantage of extensive temporal integration. Thus rod responses determine the trade-off between sensitivity (long integration time) and temporal precision (short reaction time) of the snapping behaviour near the absolute seeing threshold [179,203]. ...
From the mid-19th century until the 1980’s, frogs and toads provided important research models for many fundamental questions in visual neuroscience. In the present century, they have been largely neglected. Yet they are animals with highly developed vision, a complex retina built on the basic vertebrate plan, an accessible brain, and an experimentally useful behavioural repertoire. They also offer a rich diversity of species and life histories on a reasonably restricted physiological and evolutionary background. We suggest that important insights may be gained from revisiting classical questions in anurans with state-of-the-art methods. At the input to the system, this especially concerns the molecular evolution of visual pigments and photoreceptors, at the output, the relation between retinal signals, brain processing and behavioural decision-making.
... Moreover, integration times tend to be longer in the dark-adapted than the light-adapted state (Laughlin and Weckström, 1993;Juusola and Hardie, 2001;Reber et al., 2015;Stöckl et al., 2016a) and also longer in nocturnal than diurnal species (Laughlin and Weckström, 1993;Frederiksen et al., 2008;Stöckl A. L. et al., 2017;Frolov and Ignatova, 2019;Donner, 2021). Extremely long integration times have been measured in nocturnal toads [1.5 s (Donner, 1989)] and in a deepsea crustacean [160 ms (Moeller and Case, 1995)]. Such long integration times lead to severe blurring of moving objects (such as predators or prey) or the visual surroundings of animals that move themselves, which makes them challenging for flying animals, or animals that need to chase fast moving prey. ...
A large proportion of animal species enjoy the benefits of being active at night, and have evolved the corresponding optical and neural adaptations to cope with the challenges of low light intensities. However, over the past century electric lighting has introduced direct and indirect light pollution into the full range of terrestrial habitats, changing nocturnal animals’ visual worlds dramatically. To understand how these changes affect nocturnal behavior, we here propose an animal-centered analysis method based on environmental imaging. This approach incorporates the sensitivity and acuity limits of individual species, arriving at predictions of photon catch relative to noise thresholds, contrast distributions, and the orientation cues nocturnal species can extract from visual scenes. This analysis relies on just a limited number of visual system parameters known for each species. By accounting for light-adaptation in our analysis, we are able to make more realistic predictions of the information animals can extract from nocturnal visual scenes under different levels of light pollution. With this analysis method, we aim to provide context for the interpretation of behavioral findings, and to allow researchers to generate specific hypotheses for the behavior of nocturnal animals in observed light-polluted scenes.
... Interestingly, at higher light intensities, where there is no need to increase the signal-to-noise ratio to enhance detection, spatial summation can function to shorten visual latency (Donner, 1989). This is because the early part of the ganglion spike response relies on the early rising phase of the photoreceptor response, which scales linearly with the number of absorbed photons. ...
At the sensitivity limit of vision, the quantal fluctuations of light and neural noise in the retina and the brain limit the detection of light signals. The challenge for vision, as for all senses, lies in separating the weakest signals from the neural noise originating within the sensory system. In this thesis, I studied sparse signal detection in the vertebrate visual system (mouse and frog) at low light levels from single retinal neurons to behavioral performance.
First, we determined the sensitivity limit of amphibian color vision at low light levels. Unlike most vertebrates, amphibians are potential dichromats even at night, with two spectrally distinct classes of rod photoreceptors: common vertebrate rods (peak sensitivity at 500 nm) and an additional class called “green rods” (peak sensitivity at 430 nm). We showed that frogs in a phototaxis experiment can distinguish blue from green down to their absolute visual threshold, meaning that they have wavelength discrimination as soon as they start seeing anything. Remarkably, the behavioral blue/green discrimination approached theoretical limits set by photon fluctuations and rod noise, highlighting the sensitivity of the system comparing signals from the two different photoreceptors. Additionally, we show that the amphibian threshold for color discrimination is task- and context-dependent, underlining that sensory discrimination is not universally driven to absolute physical limits, but depends on evolutionary trade-offs and flexible brain states.
In the second paper, we studied the impact of the circadian rhythm on the sensitivity limit of mouse vision. The retina has its own intrinsic circadian rhythms, which has led to the hypothesis that the sensitivity limit of vision would be under circadian control. We used a simple photon detection task, which allowed us to link well-defined retinal output signals to visually guided behavior. We found that mice have strikingly better performance in the visual task at night, so that they can reliably detect 10-fold dimmer light in the night than in the day. Interestingly, and contrary to previous hypotheses, this sensitivity difference did not arise in the retina, as assessed by spike recordings from retinal ganglion cells. Instead, mice utilize a more efficient search strategy in the task during the night. They are even able to apply the more efficient strategy at day once they have first performed the task during the night. Measured differences in search strategy explain only part of the day/night difference, however. We hypothesize that in addition there are diurnal changes in the state of brain circuits reading out the retinal input and making decisions.
In the third paper, we determined the sensitivity limit of decrement (shadow) detection of mouse vision. Compared with the question of ultimate limit for detecting light, the question of sensitivity limits for detecting light decrements (negative contrast) has been remarkably neglected. We recorded the OFF responses of the most sensitive retinal ganglion cells at dim background light levels and correlated the thresholds to visually guided behavior in tightly matched conditions. We show that compared with an ideal- observer model most of the losses happen in the retina and remarkably, the behavioral performance is very close to an optimal read-out of the retinal ganglion cells.
I have shown across visual tasks and in two different species how closely behavior in specific conditions can approach the performance limit set by physical constraints, rejecting noise and making use of every available photon. However, the actual performance strongly depends on the behavioral context and relevance of the task and state of the brain.
... The smooth curves are model functions calculated on the assumptions that (1) the first spike occurs when the rod response linearly summed over the RF has reached a criterion amplitude, plus a constant 'transmission delay'; (2) the initial spike frequency is determined by the steepness of the leading edge of the summed rod response over a short interval after that. After Donner (1989). ...
Time is largely a hidden variable in vision. It is the condition for seeing interesting things such as spatial forms and patterns, colours and movements in the external world, and yet is not meant to be noticed in itself. Temporal aspects of visual processing have received comparatively little attention in research. Temporal properties have been made explicit mainly in measurements of resolution and integration in simple tasks such as detection of spatially homogeneous flicker or light pulses of varying duration. Only through a mechanistic understanding of their basis in retinal photoreceptors and circuits can such measures guide modelling of natural vision in different species and illuminate functional and evolutionary trade-offs. Temporal vision research would benefit from bridging traditions that speak different languages. Towards that goal, I here review studies from the fields of human psychophysics, retinal physiology and neuroethology, with a focus on fundamental constraints set by early vision.
... In addition, since the luminance was the only attribute along the pupillary reflex pathway in this study, the latency of conduction along the pupillary reflex pathway was constant. In contrast, the latency of retinal cells, cone, rod, and intrinsically photosensitive retinal ganglion cells (ipRGC) containing melanopsin is dependent on the stimulus luminance and size (Donner, 1989;Wolpert, Miall, Cumming, & Boniface, 1993;Williams & Lit, 1983;Kelbsch et al., 2019), indicating that CFD, which is the product of stimulus area by luminance, affects PL. Accordingly, PL could be depicted as a function of CFD. ...
Previous studies show that the amplitude of pupillary light response (PLR) depends on the corneal flux density (CFD), which is the product of stimulus area by luminance. However, the contribution of CFD has been investigated only when the stimulus was centered on the fovea, whereas perceived luminance to pupillary response would reduce with stimulus eccentricity. Additionally, it has been shown recently that attentional state modulates pupillary response. In this study, we aimed to clarify the complete mechanisms of PLR by manipulating the stimulus size, eccentricity, luminance, and the participants’ attentional states. We focused on four indices to examine PLR, that is, pupillary latency (PL), maximum constriction velocity (MCV), maximum constriction (MC), and mean pupil change (MPC). Results showed that PL was a function of CFD, whereas MCV, MC, and MPC were functions of both CFD and stimulus eccentricity. Furthermore, the magnitude of effect due to stimulus eccentricity for MCV and MC was different from that for MPC. These results provided new evidence that the different processing systems in PLR existed.
... In addition, since the luminance was the only attribute along the pupillary reflex pathway in this study, the latency of conduction along the pupillary reflex pathway was constant. In contrast, the latency of retinal cells, cone, rod, and intrinsically photosensitive retinal ganglion cells (ipRGC) containing melanopsin is dependent on the stimulus luminance and size (Donner, 1989;Wolpert, Miall, Cumming, & Boniface, 1993;Williams & Lit, 1983;Kelbsch et al., 2019), indicating that CFD, which is the product of stimulus area by luminance, affects PL. Accordingly, PL could be depicted as a function of CFD. ...
To process the motion of objects, humans need to consider information about up-down direction as obtained through various cues such as the gravity direction in the environment, visual polarity, and body direction. This study investigates the effects of up-down direction, as obtained from these cues on motion perception, with a focus on acceleration perception. We presented the participants with moving objects that had various acceleration speeds and measured the physical acceleration to be perceived as constant velocity. We examined the effect of the up-down direction from the visual polarity by changing the relationship between the up-down direction indicated by the gravity direction cue and the up-down direction indicated by visual polarity by manipulating the posture of the observer. The results showed that the up-down direction received by the gravity affected motion perception. Moreover, the up-down direction indicated by the visual polarity affected motion perception when the observer's body direction and the physical gravity direction were different. On the other hand, up-down direction indicated by the visual polarity did not affect motion perception when the body direction coincides with physical gravity direction. Overall, the results suggest that the up-down directions indicated by the gravity, visual polarity, and body direction are integrated non-linearly in the perceived acceleration of visual motion.
... An equally important but less commonly appreciated advantage of increased sensitivity is the significant gain in visual reaction speed, which may be as important as increased visual range when it comes to eating or being eaten. Visual latency decreases very sharply with increasing contrast just above threshold (Donner 1989;Djupsund et al. 1996;Donner & Fagerholm 2003). Even when both members of a predator-prey pair can detect each other, the member seeing the other at a higher contrast relative to its own threshold will have an advantage in the speed of vision. ...
The photoreceptors and eyes of four fish species commonly cohabiting Fennoscandian lakes with different light transmission properties were compared: pikeperch Sander lucioperca, pike Esox lucius, perch Perca fluviatilis and roach Rutilus rutilus. Each species was represented by individuals from a clear (greenish) and a humic (dark brown) lake in southern Finland: Lake Vesijärvi (LV; peak transmission around 570 nm) and Lake Tuusulanjärvi (LT; peak transmission around 630 nm). In the autumn, all species had almost purely A2‐based visual pigments. Rod absorption spectra peaked at c.526 nm (S. lucioperca), c. 533 nm (E. lucius) and c. 540 nm (P. fluviatilis and R. rutilus), with no differences between the lakes. Esox lucius rods had remarkably long outer segments, 1.5–2.8‐fold longer than those of the other species. All species possessed middle‐wavelength‐sensitive (MWS) and long‐wavelength‐sensitive (LWS) cone pigments in single, twin or double cones. Rutilus rutilus also had two types of short‐wavelength sensitive (SWS) cones: UV‐sensitive [SWS1] and blue‐sensitive (SWS2) cones, although in the samples from LT no UV cones were found. No other within‐species differences in photoreceptor cell complements, absorption spectra or morphologies were found between the lakes. However, E. lucius eyes had a significantly lower focal ratio in LT compared with LV, enhancing sensitivity at the expense of acuity in the dark‐brown lake. Comparing species, S. lucioperca was estimated to have the highest visual sensitivity, at least two times higher than similar‐sized E. lucius, thanks to the large relative size of the eye (pupil) and the presence of a reflecting tapetum behind the retina. High absolute sensitivity will give a competitive edge also in terms of short reaction times and long visual range.
... While the photoreceptor integration time in the dark-adapted state in many nocturnal insects is longer than in the light-adapted state and also longer than in many of their diurnal relatives, it is still relatively fast at about 40 to 50 ms. Very long integration times have been measured in nocturnal toads (1.5s (Donner, 1989)) and in a deep-sea crustacean (160 ms (Moeller & Case, 1995)). Such long integration times lead to severe blurring of a moving object (similar to the effects of temporal summation, see Fig. 7), which makes them unsuitable to flying animals, or animals that need to chase fast moving prey. ...
... Ganglion cells receive photoreceptor inputs via a chain of interneurons: the bipolar cells, horizontal cells, and amacrine cells ( Fig. 10.1). Despite these interneurons, the temporal characteristics (the integration time) of a ganglion cell seems to be determined by the kinetics of the photoreceptors that for the moment are driving the cell, namely, the cones at daylight, or the rods at low light (for rods, see Donner, 1989a). ...
This chapter discusses the anatomy and physiology of the retina in aquatic tetrapods. It focuses on the input and output stages of retinal processing: the photoreceptors and the ganglion cells. It describes the characteristics of rod and cone visual pigments, photoreceptor sensitivity, and the importance of temporal summation of photoreceptor signals. It also discusses ganglion cell densities and topographies, and examines how ganglion cells receive photoreceptor inputs and determine anatomical visual acuity.
... Temporal summation, in the form of decreased response speed, has been measured in the photoreceptors of nocturnal vertebrates and insects, and is found to be considerably slower than in diurnal species (Dubs, 1981;Donner, 1989;Moeller and Case, 1995). In vertebrates, spatial summation is initially carried out in the peripheral ganglion cells of the retina, pooling signals from up to hundreds of individual photoreceptors (Hughes, 1977). ...
Animals use vision over a wide range of light intensities, from dim starlight to bright sunshine. For animals active in very dim light the visual system is challenged by several sources of visual noise. Adaptations in the eyes, as well as in the neural circuitry, have evolved to suppress the noise and enhance the visual signal, thereby improving vision in dim light. Among neural adaptations, spatial summation of visual signals from neighboring processing units is suggested to increase the reliability of signal detection and thus visual sensitivity. In insects, the likely neural candidates for carrying out spatial summation are the lamina monopolar cells (LMCs) of the first visual processing area of the insect brain (the lamina). We have classified LMCs in three species of hawkmoths having considerably different activity periods but very similar ecology - the diurnal Macroglossum stellatarum, the nocturnal Deilephila elpenor and the crepuscular-nocturnal Manduca sexta. Using this classification, we investigated the anatomical adaptations of hawkmoth LMCs suited for spatial summation. We found that specific types of LMCs have dendrites extending to significantly more neighboring cartridges in the two nocturnal and crepuscular species than in the diurnal species, making these LMC types strong candidates for spatial summation. Moreover, while the absolute number of cartridges visited by the LMCs differed between the two dim-light species, their dendritic extents were very similar in terms of visual angle, possibly indicating a limiting spatial acuity. Interestingly, the overall size of the lamina neuropil did not correlate with the size of its LMCs. This article is protected by copyright. All rights reserved.
© 2015 Wiley Periodicals, Inc.
A total of 34 individual brightness functions were measured for 18 observers by two different methods. In one method the observer
set various luminance levels of a white target and assigned numbers proportional to the apparent brightness of the levels
set. In the other method the observer adjusted the loudness of a white noise and the luminance of a white target in order
to achieve a series of cross-modality matches between loudness and brightness. Both methods gave good approximations to power
functions, showing that the psychophysical power law holds for the individual perceiver.
The time-course of light-induced potential changes was measured transretinally as a function of flash intensity in the eyecup preparation of dark-adapted Bufo marinus. The electroretinogram (ERG) as indexed by the peak amplitude of the a-wave. and the latency of the a-wave onset yielded action spectra matching that of the red rod pigment (λmax at 502 nm) at low intensities near threshold, but at higher intensities there was evidence of intrusion from the single and principal short-latency, longwave cones (λmax at 575 nm). With uniform illumination of the retina with monochromatic (λ = 502 nm) flashes to isolate the red rod system over a range of 5 log units from threshold to saturation of the a-wave. the latency of the b-wave onset and the latency of the a-wave peak varied linearly with the latency of the a-wave onset. All three measures could be described by linear functions of the inverse cube root of flash intensity, or equivalently by the equation: tmin/t=I1/3(I1/3 + σ1/3). For the mass receptor potential (MRP) isolated with excess Mg2+. the initial segment and functional form of the time-course did not differ from that obtained for the a-wave at the same saturating intensity. The results are consistent with the hypothesis that the dominant mechanism controlling visual latency lies in the photoreceptors and that subsequent proximal transformations of the time-course are linear.
A power function, not a log function, describes the operating characteristic of a sensory system.
1. Outer segments of individual rods in the retina of the toad, Bufo marinus, were drawn into a glass pipette to record the membrane current. 2. Light flashes evoked transient outward currents. The peak response amplitude was related to flash intensity by a Michaelis equation with half-saturating intensity about 1 photon mum-2. 3. The saturating response amplitude ranged up to 27 pA and corresponded closely to complete suppression of the steady inward current present in darkness. 4. For a given cell the saturating response amplitude varied linearly with the length of outer segment within the pipette. This is consistent with a uniform density of light-sensitive channels and negligible gradient of membrane potential along the outer segment. 5. Responses to bright flashes never showed the relaxation from an initial peak seen previously in intracellular voltage recordings, suggesting that the conductance change responsible for the relaxation does not occur in the outer segment. 6. Responses to local illumination of only the recorded outer segment were very similar to those obtained with diffuse light at the same intensity, indicating that peripheral rods made little contribution to the responses. 7. The spectral sensitivity of 'red' rods was consistent with a retinal1-based pigment with lambda max = 498 +/- 2 nm. 8. The kinetics of the response were consistent with four stages of delay affecting action of the internal transmitter. Responses were faster at the basal end of the outer segment than at the distal tip. 9. Background light reduced the sensitivity to a superposed dim test flash and shortened the time course of the response, indicating that adapting light modifies the kinetics and gain of the transduction mechanism within the outer segment. 10. Responses to dim lights exhibited pronounced fluctuations which are attributed in the succeeding paper (Baylor, Lamb & Yau, 1979) to the quantal nature of light.
The membrane current of single rod outer segments in pieces of isolated toad retina was recorded with a glass suction electrode. Light evoked a slow net outward photocurrent consisting of a reduction in the steady inward dark current. In very dim light, the photocurrent broke up into discrete shot effects with a rounded shape and an amplitude of about 1 pA. These events were shown to result from photoisomerization of single rhodopsin molecules. The current in darkness showed fluctuations consisting of (a) discrete events apparently resulting from thermal isomerization of rhodopsin molecules, and (b) smaller amplitude shot effects shaped by two of the four rate processes of the light response.
1. Synaptic transfer between the retinal input and output was studied in turtle eyecups by injecting rectangular current pulses into a single cone or rod while recording externally from a ganglion cell.
2. When a receptor was activated with weak steps of polarizing current, the probability of obtaining a ganglion cell impulse rose after an S‐shaped delay to a peak at about 0·1 sec and then declined. This suggests that the transmission chain behaves like an electrical band‐pass filter containing delay and differentiating elements.
3. To further characterize the kinetics of excitation in the subthreshold region, the duration and polarity of the polarizing current pulses were varied while determining the magnitude of the threshold current and the delay to the ganglion cell impulses. The results of these experiments were described with linear models which assume that synaptic transfer occurs over a cascade of first‐order delay stages and a single differentiating stage.
4. The pathways which relay off responses to light from rods and red‐sensitive cones were formally similar, but the time scale in the rod path was several times slower. The path carrying off responses from the red‐sensitive cones was faster than the on path. These kinetic differences indicate that independent pathways mediate each of the three categories of response and suggest that the kinetics of each path are ‘matched’ to the input signals generated by light.
5. The strength—latency relations for the responses of on‐centre ganglion cells to flashes and steps of light were approximately predicted from the description of synaptic transfer developed here and the description of visual transduction in red‐sensitive cones from a previous study.
6. It is suggested that the retinal paths have kinetics which might be useful in transmitting light‐evoked signals whilst attenuating noise present near the input.
1. Aspartate-isolated photoresponses of the red rods to flashes and steps of light have been recorded, both in the presence of and without background lights of varying strength. 2. The results are interpreted in terms of a model of rod outer segment adaptation, where the three model parameters correspond to the adaptation processes associated with the transmitter release, the transmitter background concentration and the plasma membrane leakage, respectively. 3. The stimulus-response function deviated somewhat from the Michaelis equation U/Umax=I/(I + IH). During light-adaptation the operating curve, the stimulus-response function plotted in a log-log diagram, retained approximately its shape while moving strongly to the right along the log intensity axis and to a lesser degree downwards (Umax-decrease). 4. The movement of the operating curve was such that the rods approximately obeyed Weber's law. In the cases of flash and step of light stimuli the movement of the operating curve was about the same. 5. When a moderate background light was turned on a large decrease of sensitivity was first observed. During a period 0-5-1 min the sensitivity increased towards the stationary value. After extinguishing the background light the dark sensitivity returned in 0-5-1 min and then a period of hypersensitivity lasting typically 1 min was observed. 6. The experimental results, as interpreted according to the model, indicate that light-adaptation decreases q, the number of transmitter molecules released by one bleached rhodopsin molecule. 7. There is probably an adaptation process also in the rod inner segment, which increases the sensitivty of the rod to transient stimuli.
The impulse discharge of ganglion cells was recorded with extracellular micro-electrodes in the excised and opened eye of the common frog, Rana temporaria. The responses of different ganglion cell types to a standard moving spot with various spot-background contrasts are described. Information about such stimulus parameters as the size and contrast as a moving object is given by different classes of ganglion cells with preferences for different stimulus features. Of 171 sustained cells with small receptive fields 29 were found directionally selective, i.e. they responded well to movements only in some directions. Experiments with double stimulus fields suggest that this selectivity is due to an amacrine cell-mediated lateral inhibition nonsymmetrically arranged around the centre of the receptive field. The dichromatic colour vision of the frog is based on partly opponent signals from yellow-sensitive cones and blue-sensitive green rods. These opponent inputs make the ganglion cells respond to blue spots moving against a yellow-green background, irrespective of the relative intensities of the two colours. When the green rods are stimulated with blue light the ganglion cells produce long "on"-responses with significantly lower impulse frequencies than the short cone-mediated responses.
1. Intracellular responses to flashes and steps of light have been recorded from the outer segment and the cell body of rods in the retina of the Bufo marinus . The identification of the origin of recorded responses has been confirmed by intracellular marking.
2. Responses to flashes delivered in darkness or superimposed on a background were analysed. Responses recorded from outer segments conform to the principle of ‘spectral univariance’. The shape of the response is not affected by enlarging the spot diameter from 150 to 1000 μm.
3. The membrane potential measured in darkness at the outer segments varied from ‐15 to ‐25 mV. Injection of steady hyperpolarizing currents increases the size of the response to light; depolarizing currents reduce the response. The mean value of the input resistance is 97 ± 30 MΩ in darkness and increases by 20‐30% during illumination.
4. The responses obtained from the cell body of rods have the same shape, time course and spectral sensitivity of those recorded at the outer segment. Injection of steady current at the cell body produces different effects than at the outer segment: hyperpolarizing currents reduce the amplitude of the response to light; depolarizing currents increase the response.
5. The experimental data are fitted according to a model similar to that used to describe the responses of turtle cones (Baylor & Hodgkin, 1974; Baylor, Hodgkin & Lamb, 1974 a, b ).
6. The model reproduces the electrical responses of the rod outer segment to a variety of stimuli: ( a ) brief flashes and steps of light in dark adapted conditions; ( b ) bright flashes superimposed on background illuminations; ( c ) pairs of flashes delivered at different time intervals. Responses to hyperpolarizing steps of current are also reproduced by the model.