Data-driven models of optimal chromatic coding in the outer retina

The functional properties of the outermost retinal circuits involved in color discrimination are not well understood. Recent experimental work on zebrafish has elucidated the in-vivo activity of photoreceptors and horizontal cells as a function of the stimulus spectrum, highlighting the appearance of chromatic-opponent signals at the first synaptic connection between cones and horizontal cells. These findings, together with the observed lack of gap junctions, suggest that the mechanism yielding early color-opponency in zebrafish is dominated by inhibitory feedback. We discuss the observed neuronal activity in the context of efficient codification of chromatic information, hypothesizing that opponent chromatic signals provide optimal codification, minimizing signal redundancy. We examine whether these functional properties are general across species by studying the dynamic properties of dichromatic and trichromatic outer retinal networks. Our findings show that dominant inhibitory feedback mechanisms provide an unambiguous codification of chromatic stimuli, whereas this property is not guaranteed in networks with strong excitatory inter-cone connections, for example via gap junctions. This provides a plausible explanation for the absence of gap junctions observed in the outermost zebrafish retinal layers. In addition, our study suggests that the simplest zebrafish-like network with dominant inhibitory feedback capable of optimally codifying chromatic information requires at least two successive inhibitory feedback layers. Finally, we contrast the chromatic codification performance of zebrafish-inspired retinal networks with networks having different opsin combinations. We find that optimal combinations lead to a chromatic codification improvement of only 13% compared with zebrafish opsins, suggesting that the zebrafish retina performs nearly optimal codification of chromatic information in its habitat. 2 Author summaryRecent experimental work has evidenced that outer neuronal circuits in the zebrafish retina use color-opponent mechanisms to codify and transmit chromatic information at the first synaptic contact between cones and horizontal cells. Inspired by these findings, we propose a data-driven model to study physiological and dynamical properties of outer retinal networks and their implications for color codification across vertebrate retinal circuits. We first study our model in a large parameter space, finding that the primary biological mechanism leading to color-opponent signals is mediated by dominant inhibitory feedback, e.g., via horizontal cell synaptic connections. In contrast, strong coupling among cones leads to ambiguous chromatic codification, undesirable in the outer retina. Then, we parameterize the model using zebrafish experimental data and quantify its chromatic codification performance. Our results suggest that trichromatic retinas with inhibitory feedback are highly efficient and capture most of the chromatic information variance typical from zebrafish environments. More specifically, a comparison among zebrafish-inspired retinal networks suggests that zebrafish retinal circuits are near-optimal chromatic codification of their natural chromatic information.