Distinct inhibitory connectivity motifs trigger distinct forms of anticipation in the retinal network

TL;DR

This paper investigates how two inhibitory connectivity motifs in the retina, feed-forward and recurrent feedback, contribute to motion anticipation, revealing distinct mechanisms and speed tuning effects that enhance real-time motion processing.

Contribution

It introduces a computational model comparing feed-forward and recurrent inhibitory motifs, showing their distinct roles in motion anticipation in the retina.

Findings

Feed-forward inhibition causes forward shift of responses via subtractive inhibition.

Recurrent feedback induces phase-shifted excitatory and inhibitory waves with divisive inhibition.

Recurrent feedback exhibits a preferred speed for motion prediction, unlike feed-forward inhibition.

Abstract

Motion is an important feature of visual scenes and retinal neuronal circuits selectively signal different motion features. It has been shown that the retina can extrapolate the position of a moving object, thereby compensating sensory transmission delays and enabling signal processing in real-time. Amacrine cells, the inhibitory interneurons of the retina, play essential roles in such computations although their precise function remain unclear. Here, we computationally explore the effect of two different inhibitory connectivity motifs on the retina's response to moving objects: feed-forward and recurrent feed-back inhibition. We show that both can account for motion anticipation with two different mechanisms. Feed-forward inhibition truncates motion responses and shifts peak responses forward via subtractive inhibition, whereas recurrent feedback coupling evokes, via divisive inhibition, excitatory and inhibitory waves with different phases that add up and shift the response peak. A key difference between the two mechanisms is how the peak response scales with the speed of a moving object. Motion prediction with feedforward circuits monotonically decreases with increasing speeds, while recurrent feedback coupling induces tuning curves that exhibit a preferred speed for which motion prediction is maximal.