Spatial extent of inputs to primate ganglion cells in natural viewing conditions

Abstract

Purpose : Responses of primate retinal ganglion cells (RGCs) to natural scenes are not well understood and are poorly predicted by models that contain a linear receptive field (RF) as the first step. Previous studies have shown that under certain conditions, RGCs in various species receive input from outside the classical RF. Here we test the spatial extent of primate RGC inputs in response to natural scenes.

Methods : Large-scale multielectrode recordings were performed in peripheral macaque retina ex vivo. ON and OFF parasol cells were identified, and their RFs were measured by reverse correlation with a white noise stimulus. Natural scenes were images from the van Hateren database, presented with jitter to simulate fixational eye movements. To test the spatial extent of parasol cell inputs, responses to natural images were compared with responses to images in which one part was replaced by the image mean (Fig.1). The RF centers of the cells were fitted with a Gaussian envelope, and distances from cells to the uniform region were normalized to the standard deviation (SD) of this envelope.

Results : Clear changes in responses were observed only when the RF significantly overlapped the uniform region (Fig. 2), specifically, if the border was within 4 SDs of the RF center (roughly the edge of the surround). However, RGCs in the uniform region showed modulated light responses, primarily at image transitions, up to 6 SDs from the border.

Conclusions : Parasol cells stimulated with natural scenes in the RF appear little affected by inputs from outside the RF. However, signals for events like image transitions may affect distant parasol cells not receiving stimulation in their RFs. Polyaxonal amacrine cells, which have long processes and show clear responses at image transitions, could produce such a signal.

This is an abstract that was submitted for the 2016 ARVO Annual Meeting, held in Seattle, Wash., May 1-5, 2016.

 

The two stimulus conditions, overlaid with the mosaic of OFF parasol RF centers for one recording. Recorded firing rates are shown for three cells: one covered by the natural image in both conditions (red), one near the border (green), and one in the uniform region (blue). Image transitions are shown in gray.

 

The mean squared error between the firing rate over time in the two conditions, normalized by the average rate, as a function of distance from the RF center to the border of the uniform region. Positive distances are above the border as shown in Fig. 1. The average measured RF profile is shown in gray, centered at 4 SDs.