Directly comparing fly and mouse visual systems reveals algorithmic similarities for motion detection

Evolution has equipped vertebrates and invertebrates with neural circuits that selectively encode visual motion. In both flies and mice, early visual circuits separately detect the movement of light and dark edges. Strong parallels have been noted between the two systems in their anatomy, circuitry, and the suite of computations that they perform. However, because their direction-selective cells have different morphologies and employ different neurotransmitters and receptors, the similarities between them exist not at the molecular level, but at the algorithmic level. While similarities in the computations performed by these circuits in mouse and fruit fly have been noted, direct experimental comparisons have been lacking. Mouse retinal physiologists and fly visual neuroscientists use largely non-overlapping sets of visual stimuli, making direct comparisons between the systems difficult. Here, we have directly compared motion encoding in these two species at the algorithmic level, using matched stimuli and focusing on a pair of analogous neurons: the mouse ON-starburst amacrine cell (ON-SAC) and Drosophila T4. Our analysis shows that the cells have similarly shaped spatiotemporal receptive fields, respond sensitively to fast spatiotemporal correlations, and are similarly tuned to sinusoidal drifting gratings, but they differ in their response to apparent motion stimuli. Interestingly, both neuron types show a response to summed sinusoids that deviates from models for motion processing in these cells, underscoring the similarities in their processing and identifying response features that remain unexplained. Flies and mice diverged many hundred million years ago, implying these similarities are likely the result of convergent evolution. This suggests strong selective pressure on motion detection to employ what may be a limited number of solutions within the anatomical and biophysical constraints of nervous systems and the statistical properties of natural visual input.