Abstract
BACKGROUND: Receptive fields of retinal neural signals of different origin can be determined from extracellular microelectrode recordings at the inner retinal surface. However, locations and types of neural processes generating the different signal components are difficult to separate and identify. We here report epiretinal receptive fields (RFs) from simultaneously recorded spikes and local electroretinograms (LERGs) using a semi-chronic multi-electrode in vivo recording technique in cats. Broadband recordings were filtered to yield LERG and multi unit as well as single unit spike signals. RFs were calculated from responses to multifocal pseudo-random spatiotemporal visual stimuli registered at the retinal surface by a 7-electrode array. RESULTS: LERGs exhibit spatially unimodal RFs always centered at the location of the electrode tip. Spike-RFs are either congruent with LERG-RFs (N = 26/61) or shifted distally (N = 35/61) but never proximally with respect to the optic disk. LERG-RFs appear at shorter latencies (11.9 ms +/- 0.5 ms, N = 18) than those of spikes (18.6 ms +/- 0.4 ms, N = 53). Furthermore, OFF-center spike-RFs precede and have shorter response rise times than ON-center spike-RFs. Our results indicate that displaced spike-RFs result from action potentials of ganglion cell axons passing the recording electrode en route to the optic disk while LERG-RFs are related to superimposed postsynaptic potentials of cells near the electrode tip. CONCLUSION: Besides contributing to the understanding of retinal function we demonstrate the caveats that come with recordings from the retinal surface, i.e., the likelihood of recordings from mixed sets of retinal neurons. Implications for the design of an epiretinal visual implant are discussed.