Topic: The development of neural coding in the zebrafish tectum, and a new statistical model for separating evoked and spontaneous activity.
Abstract: During early life, the neural code must develop to appropriately transform sensory inputs into behavioural outputs. However little is known about how refinements in neural representations over development directly impact behaviour. In zebrafish larvae we show that visually-driven hunting behaviour improves from 4 to 15 days post-fertilization, becoming faster and more accurate. During the same period population activity in the optic tectum refines, leading to improved decoding and information transmission of spatial position, particularly in the representation of the frontal visual field. Remarkably, individual differences in decoding can predict each fish’s hunting success. Together these results show how the neural codes required to subserve a natural behaviour emerge during development.
These data display the ubiquitous issue that the pattern of neural activity evoked by a stimulus can be substantially affected by ongoing spontaneous activity. Uncovering the interplay between stimulus-evoked and spontaneous activity requires the ability to reliably separate these two components. This is challenging, however, as the internal factors that give rise to spontaneous activity typically cannot be directly measured. To address this we have recently developed a new latent variable model CILVA (Calcium Imaging Latent Variable Analysis) to decouple the components of calcium imaging data due to evoked activity from those driven by structured spontaneous activity. We use Bayesian methods to fit the model to data and identify low dimensional structure underlying spontaneous activity. Applying CILVA to experimental allows us to characterise how neurons are differentially driven by external stimuli, latent internal factors, and their interaction.