The brain computes using neurons that each integrate thousands of inputs to determine their firing output. We are developing tools to record those input patterns at high speed, to study computations that occur in individual neurons during behavior.
Neurons can have thousands of synapses distributed in three dimensions along their dendrites. The computations that neurons perform are determined by patterns of synaptic activity in space and time. Much like currents in transistors, subthreshold synaptic inputs can interact, allowing neuronal dendrites to perform nonlinear integration. However, we know very little about how different kinds of neurons combine their inputs to perform behaviorally-relevant computations. To better understand how the brain processes information, we need tools to directly read out intricate three-dimensional patterns of synaptic activity in behaving animals, so that we can infer how neurons transform inputs into outputs.
Our lab is making tools to help tackle this challenge on three fronts. We are developing microscopes that record activity at high speed, fluorescent indicators that report subthreshold inputs, and computational methods that infer neuronal transformations from measured inputs and outputs. Our ultimate goal is to simultaneously record all synaptic inputs to a cortical neuron in a behaving mouse and use those data to infer how that neuron transforms information.