A neural circuit for stereopsis in the fly

Abstract

Stereopsis involves combining different views from the two eyes to infer depth, an ability that has intrigued neuroscientists for almost two centuries (Wheatstone, 1838). It enables insects such as the praying mantis to estimate the distance of prey (Nityananda et al., 2018), or primates to fixate objects at different depths (Cumming and Judge, 1986; Rashbass and Westheimer, 1961). While theoretical models for stereopsis have been proposed based on neural activity recordings (Marr and Poggio, 1976; Grossberg, 1983; Ohzawa et al., 1990), we do not have a detailed circuit mechanism for stereopsis founded on neural wiring. Here we achieve this in the fly, Drosophila melanogaster, by analyzing two whole-brain connectomes (Dorkenwald et al., 2024; Berg et al., 2026). Flies have sufficient overlap between the eyes to utilize binocular vision (Kemppainen et al., 2022); however, a circuit that performs stereopsis has not been uncovered in this model system. We show that a cell type that connects the two optic lobes called LC14 (Otsuna and Ito, 2006) forms a recurrent circuit across the brain that is capable of depth estimation. By building a connectome-constrained model, we demonstrate that recurrence is critical for binocular distance tuning in this circuit. Tuning is optimal for the size and distances of social targets (other flies), and we uncover a pathway within the brain specific for binocular vision computations. Silencing LC14 neurons renders this downstream circuitry sensitive to only monocular information and alters distance-dependent courtship behaviors, while activation of LC14 neurons causes flies to slow in response to an illusory object in their binocular visual field. In sum, we predict function from structure, uncovering how the fly brain computes distance from binocular cues, and identify a role for binocular vision in patterning social behavior.