Classification of object size in retinotectal microcircuits

A principal task of the visual system is to detect and classify moving objects in the visual environment. Information about the size of an object is critical for selecting appropriate behavioral responses. Object size is encoded in retinal ganglion cell (RGC) activity. Little is known, however, about how inputs from the multitude of RGC subtypes are distributed to higher visual centers and how information is combined from these feature-selective inputs.

Here we show that in the zebrafish optic tectum, prey- or predator-like moving targets evoke activity in distinct groups of RGC fibers dependent on target size, demonstrating a retinal origin of tectal size classification. Small-size-selective retinal inputs are relatively more frequent in the most superficial layer of the tectal neuropil, whereas large-size-selective inputs predominate in deeper layers. Monostratified superficial interneurons (SINs) process large-size- and small-size-selective signals dependent on their dendritic target layer, consistent with the retinal input organization. Further downstream, small- and large-sized objects are encoded in population activity of separate sets of tectal neurons.

Ethologically relevant size classes are preferentially processed in different layers of the tectal neuropil. The tectum categorizes visual targets on the basis of retinally computed size information, suggesting a critical role in visually guided response selection.