Introduction

Retinas are geared to function as the first cell layers of the visual system by several sorts of signal generation and processing. In addition to transducing light and limiting light sensitivity, photoreceptors help reduce noise, differentially distribute signals to different bipolar cells, and adapt to changes in mean luminance (Sterling, 1998). Before driving retinal ganglion cells to “spike” (i.e., generate action potentials), interneurons help determine the spatial, temporal, and chromatic properties that produce changes in ganglion cell spiking at any moment, and help adjust the sensitivity and response kinetics of cells when the mean, range, and spatial distribution of light intensities change (Meister and Berry, 1999). Finally, the outcome of these processes is enriched by the variety of cellular teams that retinas devote to handling these tasks and, as described in other chapters of this book, this variety arises at least in part from differences in the combinations of bipolar and amacrine cells that converge onto ganglion cells, the synaptic arrangements and transmitter systems that make up the outer and inner plexiform layers, and the transfer and dynamic properties of these synapses.

In the midst of this, gangion cells may seem to simply be on stand-by, generating a spike anytime some fixed threshold is reached. However, certain responses of the retina also entail changes in signal flow through synapses and single cells. The best known of these include shifts in the balance of rod and cone inputs to shared pathways, modulation of neurotransmitter-gated and voltage-gated ion channels, and changes in coupling at gap junctions (Akopian and Witkovsky, 2002; Barlow, 2001; Dowling, 1991; Vaney, 1999; Wässle and Boycott, 1991; Witkovsky and Dearry, 1991). Because some modulators have recently been found to regulate ion channels in ganglion cells, retinas seem set up to generate their outputs in part because stimuli change the synaptic inputs that drive ganglion cells, and in part because stimuli change the response of ganglion cells to their inputs. This chapter addresses two questions about how ganglion cells fit into an ensemble like this to form the retinal output. First, do all ganglion cells generate spikes in the same way? Second, do individual ganglion cells produce spikes the same way under all lighting conditions?

One reason for considering these possibilities is that they imply that retinas could be organized in different ways to respond to their inputs. A fixed mechanism of ganglion cell “excitability” (i.e., spike generation) would suffice to send the retina's output off to remote targets without it degrading along the way, but this would leave information-processing duties in the retina up to the outer and inner nuclear layers. Allowing different ganglion cells to generate spikes at different frequencies, at different moments, or in different temporal patterns could create different outputs for identical inputs, relieve other retinal cells from shaping these spiking patterns, or ensure that critical processing that started in the distal retina was maintained in the retinal output. Equipping individual ganglion cells with spiking mechanisms that change depending on the intensity, distribution, or other attributes of light could enable ganglion cells to modulate the retina's output, and this could occur independently of or together with changes in the light responses of other retinal neurons.

A second reason for considering ganglion cell excitability is that it could factor into the range of inputs these cells respond to. For example, ganglion cell responses, measured in spike firing rate (i.e., spike “frequency”), saturate within a “dynamic range” of 2.5 log units (Barlow and Levick, 1969; Diamond and Copenhagen, 1995; Enroth-Cugell and Shapley, 1973; Green and Powers, 1982; Sakmann and Creutzfeldt, 1969; Thibos and Werblin, 1978), yet photon fluxes encountered under starlight and sunlight can differ by up to 10 log units (Rodieck, 1998). For ganglion cells to respond over broad ranges of stimulus values, retinas with fixed spike-generation mechanisms in all ganglion cells, or with different mechanisms in different cells, have at least three sorts of options. One is “range fractionation,” the use of different cells to respond to different portions of a parameter's value range (cf. Albrecht and Hamilton, 1982; Ohzawa et al., 1985). A second way is “receptive field surround inhibition,” the reduction of responses to localized “center” illumination by light that falls within a slightly broader “surround” area (Barlow et al., 1957; Barlow and Levick, 1969; Donner, 1981; Enroth-Cugell and Lennie, 1975). A third would be “adaptation” or “gain control”, which adjust the response that cells produce per unit of input (Shapley, 1997; Shapley and Enroth-Cugell, 1984). Although none of these tactics necessarily entail differences or changes in ganglion cell excitability, differences in the range of inputs that ganglion cells respond to or the range of outputs that ganglion cells generate could contribute to range fractionation, and changes in excitability could result from changes in either local or broad-field illumination.

One approach to take here would be to proceed through descriptions of light stimuli, cell types, light responses, mechanisms, and models. Unfortunately, although a number of different models of ganglion cell light responses have been built from comparisons of stimuli and responses, surprisingly few biophysical and intracellular signaling mechanisms that allow these models to work have been established by intracellular recording, imaging, or experimental manipulations. This chapter will therefore start by describing simple light responses of ganglion cells, those of dark-adapted cells to flashes of intermediate intensity. A number of cellular properties that have been found in ganglion cells by patch-clamp and immunocytochemical methods will then be reviewed, and related, where possible, to responses of light-adapted ganglion cells to light flashes and to randomly fluctuating intensities. Two of the main ideas considered here are that ganglion cell spike generation is not an “integrate-and-fire” process that can be described by voltage-gated ion current amplitudes, kinetics, and voltage sensitivities that are identical in all cells and constant under all conditions, and that changes in ganglion cell excitability help adjust the retinal output as the retinal input changes.