Neural computation from and beyond connectivity

Abstract

To generate adaptive behavior in a dynamic world, the brain must represent information in the activity of its constituent cells and transform this information through the communication of these cells into appropriate motor and autonomic output. This process can be conceptualized as a series of computations, and my dissertation investigates how the brain implements these computations. I first argue for a characterization of computational substrates in the brain into two broad categories: ‘connectomic’ substrates arising from precise neural connectivity, and ‘extra-connectomic’ substrates beyond connectivity. I focus on a specific behavior, the optomotor response. The larval zebrafish swims in the direction of the whole-field optic flow to stabilize its position, but quickly stops if its swims do not result in any visual feedback. This dissertation investigates the implementation of the optomotor response and its inhibition by futility from sensory input to motor output. First, I helped generate a connectome, or map of all synaptic connections, of a larval zebrafish brain together with functional imaging of the same brain. Using this functional connectomics dataset, I reconstructed the circuits transforming directional whole-field motion into swim commands and those linking a lack of swim-related visual feedback to activation of noradrenergic neurons in the hindbrain. Next, going beyond the connectome, I identified an extracellular biochemical signaling pathway through which astroglia activate inhibitory neurons. Finally, I return to the connectome to investigate information flow downstream of inhibitory neuron activation. Altogether, this dissertation provides a framework through which insights from and beyond synaptic connectivity are combined for investigating circuit function in the vertebrate brain.