2022BBSRC-NSF/BIO Generating New Network Analysis Tools for Elucidating the Functional Logic of 3D Vision Circuits of the Drosophila Brain

Human eyes constantly exert small-scale motions called microsaccades. The functional role of microsaccades in perceiving the visual world remains a mystery. The lack of suitable tools and methodologies has limited our understanding of how animals see the 3-dimensional visual world and control their movements. The primate eyes and brain have hindered simultaneous live imaging of large-scale clusters of neurons at cellular resolution.

Conversely, approaches using small-brain model organisms, such as the fruit fly (Drosophila) and zebrafish, have focussed on assigning motion detection and colour discrimination tasks to single neurons and circuits. Recently, it has been discovered that the photoreceptors in the retina of the fruit fly are also subject to light-induced and muscle-induced retinal movements such as microsaccades. It was shown in the fruit fly that light-induced microsaccades play a key role in depth perception of stereo vision.

In this proposal, we aim to generate novel tools to elucidate how Drosophila brain circuits work dynamically at subcellular resolution in the biochemical domain. These tools integrate activity recordings in the living eye and biological brain circuits with graph analysis, multiscale modelling, active visual sampling, and connectomics/synaptomics models of computation. Building on this integration, we will investigate how active sampling in the retina supports Drosophila visual networks to encode 3-dimensional object information. Our novel multidisciplinary approach will generate new open-source software tools that will be invaluable for the neuroscience community exploring how stereoscopic perception of 3D objects emerges from the complex synaptic interactions of the eyes' and brain's visual circuits.

Individually and collectively, these tools will potentially have broad impact well beyond the immediate goal of elucidating the functional logic of 3D vision circuits in the Drosophila.

We recently demonstrated that Drosophila exhibits stereopsis. However, the functional logic of stereopsis in its visual system remains elusive. Our goal here is to generate new experimental and computational network analysis tools for studying the role active sampling induced by retinal movements plays in the processing of 3D object information in the early visual system.

In traditional neuroscience, the neuron is the main computational building block, while the role of synapses is to interconnect neurons. Our key hypothesis here takes this view "upside down" by arguing that most computations in the brain occur naturally in the biochemical domain and not in the electrical domain. Consequently, we approach modelling 3D visual processing on two radically different levels of abstraction. The 1st, called the NeuroCentric model, embraces the traditional level of abstraction of neurons. The 2nd, called the SynaptoCentric model, operates at the level of abstraction of synapses, where the primary objects of processing are groups of synapses communicating using neuron "wires". This alternative computing paradigm shifts processing from the electrical to the biochemical (molecular) domain.

We propose to test the theory of superresolution stereopsis and the functional logic of fly 3D vision circuits by designing a new powerful infrastructure integrating bioinformatics and neuroinformatics tools operating on fruit fly brain datasets. This plan contains four central aims: 1) elucidate how active sampling and adaptive synaptic functions support 3D vision, 2) illuminate how neural pathways dynamically distribute feature information in support of 3D perception, 3) uncover the NeuroCentric and SynaptoCentric principles of 3D phase information processing in response to microsaccadic retinal movements, & 4) create a programmable ontology platform supporting NeuroCentric and SynaptoCentric analysis tools for elucidating the functional logic of 3D Vision circuits of the fly brain.