A population code for spatial representation in the larval zebrafish telencephalon

Abstract

The vertebrate telencephalon is the site of complex cognitive processes, such as spatial cognition. The larval zebrafish telencephalon is a compact circuit of only [~]10,000 neurons that contains potentially homologous structures to the mammalian basal ganglia and limbic system (e.g., the hippocampus). However, despite long-standing evidence that spatial navigation and learning in zebrafish requires an intact telencephalon, cells believed to underlie spatial cognition in the mammalian hippocampus (e.g., place cells) have yet to be established in any fish species. Using a tracking microscope to image brain-wide activity at cellular resolution in freely swimming larval zebrafish, we compute the spatial information of neurons throughout the zebrafish brain. Strikingly, in every animal we recorded, cells with the highest spatial specificity are enriched in the zebrafish telencephalon. These cells form a population code of space, from which we can decode the animals spatial location across time. By continuous recording of population-level activity, we find that the activity manifold of place cells gradually untangles over time. Through systematic manipulation of allothetic and idiothetic cues, we demonstrate that place cells in the zebrafish telencephalon integrate multiple sources of information. By analysis of neighborhood distance between cells across environments, we find that the spatial representation in the zebrafish telencephalon partially generalizes across environments, suggesting that preconfigured network states may have been a feature of spatial computation that emerged early in vertebrate evolution.