Jon Gottesman

A power law for perceived contrast in human vision

Abstract The dependence of perceived contrast on stimulus contrast of sinewave gratings was measured by the method of magnitude estimation. The resulting perceived contrast functions are well described by threshold-corrected power functions with exponents near 0.7. The exponents are insensitive to changes in mean luminance from 10–340 cd/m 2 and to changes in spatial frequency from 0.25 to 12 c/deg. The exponents are also insensitive to a change in the range of grating contrasts from 1–2 log units. However, the distribution of contrast levels within the range produces small, but predictable, effects. Several factors are identified that may account for discrepancies in previous measurements of perceived contrast functions.

Symmetry and constancy in the perception of negative and positive luminance contrast

The perception of suprathreshold luminance contrast was investigated by forced-choice psychophysical procedures that were designed to define contrast equivalence relations. Observers compared the perceived contrast of rectangular bars that were presented for 500 msec at 3.9 deg on opposite sides of the fovea. The results show a nearly symmetrical relation between the perception of negative and positive contrast that is largely invariant over four decades of background luminance. Thus, for any fixed background luminance, equal absolute contrasts evoke approximately equal perceived contrasts. Symmetry also held with variations in the width, the eccentricity, and the focus of the bars. Symmetry was investigated further by determining equivalent contrast relations for negative contrasts as a function of background luminance and by contrast scaling. These results show evidence for nearly perfect contrast constancy for targets of low to moderate contrast and departures form constancy for high-contrast targets. These new findings on negative contrast, symmetry, and contrast constancy are discussed in relation to underlying mechanisms for contrast perception and classic experiments on brightness and lightness constancy.

A computational model of the ribbon synapse.

A model of the ribbon synapse was developed to replicate both pre- and postsynaptic functions of this glutamatergic juncture. The presynaptic portion of the model is rich in anatomical and physiological detail and includes multiple release sites for each ribbon based on anatomical studies of presynaptic terminals, presynaptic voltage at the terminal, the activation of voltage-gated calcium channels and a calcium-dependent release mechanism whose rate varies as a function of the calcium concentration that is monitored at two different sites which control both an ultrafast, docked pool of vesicles and a release ready pool of tethered vesicles. The postsynaptic portion of the program models diffusion of glutamate and the physiological properties of glutamatergic neurotransmission in target cells. We demonstrate the behavior of the model using the retinal bipolar cell to ganglion cell ribbon synapse. The model was constrained by the anatomy of salamander bipolar terminals based on the ultrastructure of these synapses and presynaptic contacts were placed onto realistic ganglion cell morphology activated by a range of ribbon synapses (46-138). These inputs could excite the cell in a manner consistent with physiological observations. This model is a comprehensive, first-generation attempt to assemble our present understanding of the ribbon synapse into a domain that permits testing our understanding of this important structure. We believe that with minor modifications of this model, it can be fine tuned for other ribbon synapses.

cellular mechanisms for color-coding in holostean retinas and the evolution of color vision

Electrophysiological recording and microspectrophotometry were used to analyze retinal function in representatives of the two surviving genera of holostean grade fish--the bowfin (Amia calva) and gars (Lepisosteus sp.). The properties of the cone photopigments, horizontal cells and ganglion cells show that these holostean retinas have cellular mechanisms for color vision which are fundamentally similar to those previously described for teleosts, turtle and mammals. These findings suggest that trichromatic receptor systems and opponent color-coding mechanisms may have evolved in primitive Neopterygii or more ancient fish, before the advent of teleosts. In conjunction with other recent data on living representatives of primitive fishes, these findings also add renewed plausibility for the view that vertebrate color vision could have taken a common origin some 400 million years ago from an ancestral aquatic jawed vertebrate.

Structure and functional connections of presynaptic terminals in the vertebrate retina revealed by activity-dependent dyes and confocal microscopy

The fluorescent dyes sulforhodamine 101 (SR 101) and FM1‐43 were used as activity‐dependent dyes (ADDs) to label presynaptic terminals in the retinas of a broad range of animals, including amphibians, mammals, fish, and turtles. The pattern of dye uptake was studied in live retinal preparations by using brightfield, fluorescence, and confocal microscopy. When bath‐applied to the retina‐eyecup, these dyes were avidly sequestered by the presynaptic terminals of virtually all rods, cones, and bipolar and amacrine cells; ganglion cell dendrites and horizontal cells lacked significant dye accumulation. Other structures stained with these dyes included pigment epithelial cells, cone outer segments, and Müller cell end‐feet. Studies of dye uptake in dark‐ and light‐adapted preparations showed significant differences in the dye accumulation pattern in the inner plexiform layer (IPL), suggesting a dynamic, light‐modulated control of endocytotic activity. Presynaptic terminals in the IPL could be segregated on the basis of volume: bipolar varicosities in the IPL were typically larger than those of amacrine cells. The combination of retrograde labeling of ganglion cells and presynaptic terminal labeling with ADDs served as the experimental preparation for three‐dimensional reconstruction of both structures, based on dual detector, confocal microscopy. Our results demonstrate a new approach for studying synaptic interactions in retinal function. These findings provide new insights into the likely number and position of functional connections from amacrine and bipolar cell terminals onto ganglion cell dendrites. J. Comp. Neurol. 437:129–155, 2001.

Response properties of C-type horizontal cells in the retina of the bowfin

Intracellular recordings were obtained from biphasic- and triphasic-type horizontal cells (C cells) in the retina of the bowfin. For steady-state responses, both cell types displayed a linear stimulus-response function for responses up to at least 20% of maximum. In the linear range, responses to red/green mixtures were well predicted from the assumption that opposed inputs combine by simple summation. Action spectra were measured in the linear range for 30 biphasic and 12 triphasic cells. Biphasic cells showed their peak hyperpolarization near 530 nm and peak depolarization near 680 nm. Triphasic cells showed peak hyperpolarization near 450 nm, peak depolarization near 570 nm and small hyperpolarizing responses to deep red flashes (greater than 670 nm). The response to deep red test flashes was reduced by chromatic backgrounds which either depolarized or hyperpolarized the cell, in contrast to past findings in carp triphasic cells. In both classes of cells, the depolarizing input mechanism had a shorter latency than the hyperpolarizing mechanism, a result not previously observed in other fish retinas. Color opponency was maintained in both classes of C cells for stimuli of small diameter. The findings in bowfin and other species suggest that both feedback and direct pathways shape the depolarizing response of C cells.

Calcium channel immunoreactivity in the salamander retina

This study reports the distribution of the α1D and α1E calcium channel subunits in the neotenous tiger salamander retina based on immunohistochemical techniques. Confocal and light microscopy were used to localize staining with fluorescently tagged antibodies to α1D and α1E in cross-sectional and flatmount preparations of retina. α1D-immunoreactivity (α1D-IR) was localized to the inner and outer plexiform layers (IPL and OPL, respectively), ganglion cell layer (GCL), and optic fiber layer. α1E-IR was found predominantly in the IPL, with scattered, weak representation in the OPL. α1E-IR was not detected in the GCL or fiber layer. These findings suggest that different α1 calcium channel proteins have distinctive distributions in retina, which may reflect their unique and different roles in retinal processing and homeostasis.

Sensory latency and reaction time: dependence on contrast polarity and early linearity in human vision

Relations between luminance contrast and reaction time were studied for foveal vision over a three-decade range of background luminance. On each background, the contrast equivalence relation between negative and positive contrast flashes conformed almost exactly to the result expected if equal luminance steps of opposite sign produce equal visual effects. The same result held for threshold detection for flashes of variable duration. Analysis of these data suggests that reaction time is triggered by the early, rising phase of an internal response and that the effective stimulus energy that triggers the response is only moderately suprathreshold. On all backgrounds the sensory latency for reaction time (L) was described reasonably well by the relation L = bS-0.67, where b is constant and S is the absolute value of the luminance step. This implies that reaction time is largely independent of contrast polarity and the background luminance. Parallels between the present results and recent intracellular work suggest that the contrast equivalence relation for reaction time is largely shaped by early linear mechanisms in cones.

N -methyl-D-aspartate receptors contribute to the baseline noise of retinal ganglion cells

Whole-cell recordings of tiger salamander ganglion cells were obtained using a superfused retinal slice preparation. Membrane current fluctuations were recorded under voltage-clamp conditions with cells usually held at -70 mV. Current fluctuations at rest (Mg2+ = 1 mM) were reduced by adding D-2-amino-7-phosphonoheptanoate (D-AP7). Resting fluctuations were increased by adding N-methyl-D-aspartate (NMDA) or by removing extracellular Mg2+. These increased fluctuations were blocked by D-AP7. Blocking NMDA receptors under control conditions also reduced a tonic inward current by -1 to -15 pA. Fluctuation analysis of current noise shows that the noise power spectrum measured in the presence of NMDA is similar to that measured under Mg(2+)-free conditions. We conclude that NMDA receptors are active in cells held at -70 mV even in the presence of 1 mM Mg2+. We believe this activation is due to the presence of endogenous glutamate in the retina. The observations of this study strongly suggest that NMDA receptors contribute to the resting noise and conductance properties of retinal ganglion cells. Our results suggest NMDA receptors are activated by an ambient level of extracellular glutamate whose source has yet to be determined.

An eyecup slice preparation for intracellular recording in vertebrate retinas

This report describes a new preparation for intracellular recording from the vertebrate retina, the eyecup slice preparation. It consists of a small (2 X 5 mm) strip cut from the posterior wall of the eye and thereby keeps the sclera, pigment epithelium and neural retina in place. Initial results are presented here for two vertebrates: the turtle, Pseudemys scripta elegans, and the toad, Bufo marinus. With conventional microscopy, the histological layers of the retina can be resolved, as well as individual photoreceptors, to provide landmarks for intracellular recording. When superfused, the eyecup slice remains in good condition for many hours and yields intracellular recordings of good quality and stability. Recordings of the light-evoked responses of cones and horizontal cells show that the slice is large enough to preserve the characteristic spatial interactions mediated by the laterally coursing neural networks of the distal retina. Recordings from rods show that full dark adaptation is achieved. Thus, photochemical dark adaptation as well as other normal cellular interactions between the neural retina and pigment epithelium can be preserved in this preparation, in contrast to isolated retinal slice preparations. The eyecup slice preparation might be particularly useful for work on mammalian retinas.