Spatial subdomains in the Optic Tectum for the encoding of visual information

Neurons across the visual system provide estimates of the visual features they encode. However, the reliability of those estimates can vary across the neuronal population. Here, we use information theory to provide a spatial map of how well neurons can distinguish ethologically-relevant visual stimuli across the entire larval zebrafish optic tectum (unknown sex), a brain region responsible for driving visually guided behaviour. We find that the ability of neurons to discriminate between stimuli is non-uniformly distributed across the tectum. Specifically, we show that information about local motion is preferentially encoded in the posterior tectum, whilst information about whole-field motion is preferentially encoded in the anterior tectum. This is achieved through two systematic changes along the anterior-posterior axis of the tectum: (i) a change in the number of neurons that discriminate between stimuli and (ii) a change in how well each neuron can discriminate between stimuli. By classifying neurons into distinct subtypes based on their response properties we uncovered a small group of neurons that are spatially localised to specific regions of the tectum and are able to discriminate between visual stimuli in a highly reliable manner. Furthermore we show these spatial biases are enhanced when using population activity to decode the visual stimuli. Our results highlight the importance of implementing information theoretic approaches to assess visual responses and provide a novel description of regional specialisation in the zebrafish optic tectum.<b>Significance Statement</b> To understand how neurons encode information about sensory stimuli it is important to establish which features they respond to and how reliably they respond. However, many studies analyse trial-averaged neuronal responses which convey no information about response reliability. We overcome this shortcoming by using an approach that considers feature selectivity along with trial-to-trial response variability allowing us to link which feature a neuron is encoding and how well it is encoded.Using this method we uncovered a high degree of spatial organisation within the zebrafish optic tectum such that different visual features are encoded preferentially within different regions. Since the tectum is topographically mapped, the spatial segregation of visual processing implies that visual objects are encoded in a location-dependent manner.