Research Vision Research Insights

Efficient information coding in the human retina

This Research Insight covers a study from the Kerschensteiner Lab. Here, we highlight how Florentina Soto, PhD and colleagues characterized the functional properties of the output cells of the retina to understand how information is coded in the human retina. 

Florentina Soto, PhD | Associate Professor, Ophthalmology and Visual Sciences

This study from the Kerschensteiner Lab, published in Neuron, marks the first functional characterization of the human retina. Florentina Soto, PhD—associate professor of ophthalmology at WashU Medicine—and colleagues performed large-scale multielectrode array recordings of retinas recovered from patients that underwent eye surgery to reveal the efficiency with which midget and parasol ganglion cells encode information in the human retina. 

Characterizing the functional responses of retinal neurons 

Midget and parasol ganglion cells are the most abundant retinal ganglion cells in the human retina. Together, their ON- and OFF-responsive subtypes, which encode information about light increments and decrements, respectively, account for approximately 80% of the signals sent from the retina to the brain. The authors reasoned that each subtype should encode minimally redundant stimulus features to limit the metabolic demand posed to the retina.  

Soto and colleagues tested this hypothesis by analyzing the visual stimuli that optimally activated each midget and parasol ganglion cell subtype. They found that the spiking responses of ON- and OFF-responsive midget and parasol ganglion cells complement one another by signaling discrete, minimally overlapping spatiotemporal domains of the visual scene. 

A flickering visual stimulus was projected onto light-sensitive photoreceptors. Electrical responses were then recorded from the retinal ganglion cells to calculate their response properties.

Their analyses revealed consistent, systematic differences between both ON- and OFF-responsive cells and between midget and parasol ganglion cells. Across both midget and parasol ganglion cells, ON-responsive cells displayed more transient responses triggered by larger visual stimuli than their OFF-responsive counterparts. Additionally, the ON-responsive cells exhibited lower spike thresholds and responded across a broader dynamic range than the OFF-responsive cells. Layered on top of these ON/OFF response asymmetries, the parasol cells displayed more transient responses with larger spatial receptive fields and lower spike thresholds than the midget cells.  

Together, these results and subsequent computational modeling experiments suggest that ON- and OFF-responsive midget and parasol ganglion cells effectively divide the spatiotemporal features of the visual scene into four discrete domains. By doing so, these ganglion cell types are sufficient to transmit complete information about the visual scene from the retina to the brain, while simultaneously minimizing the spikes – and therefore energy – required to do so. 

The visual response system consisting of ON- and OFF-responsive midget and parasol cells is effectively optimized to minimize energy expenditure (red arrows).