Project Summary/Abstract Satisfying explanations of the physiological function of a tissue, which help guide medical interventions, frame that function in terms of the inputs of component cells and an algorithm for how those cells transform their inputs into outputs. Brain functions have so far eluded such mechanistic explanation, in part because 1) the component cells – neurons – each combine up to thousands of synaptic inputs to generate their output, and because 2) it is difficult to determine how any given neuron contributes to the function of the brain as a whole. As a result, we do not have explanations in the above terms for mammalian brain circuits, nor are we able to measure the input-output operations of even a single neuron in the mammalian brain. Addressing the above challenges will aid design of medical interventions in the brain, especially of therapeutic devices that must directly interface with neurons – so-called brain-machine interfaces (BMIs). I will address the first challenge by using sensitive new genetically encoded neurotransmitter indicators (GETIs) and a novel high-bandwidth in vivo microscope to simultaneously record the activity of thousands of synaptic inputs and outputs within individual neurons in the cortex of behaving mice. I will build on my recent work developing a high-sensitivity GETI for glutamate by developing a spectrally-compatible pair of GETIs for glutamate and GABA. I will complete the development of the 2nd generation Scanned Line Projection Microscope (SLAP2), an in vivo microscope that will accurately and efficiently record from thousands of synapses in 3D at >100 Hz. Together these tools will make it possible to directly see, at high speed, the precise timing and location of myriad neurotransmitter inputs to a neuron, observe how those inputs line up to drive firing, and watch in real-time as inputs change with learning. To overcome the second challenge and enable reliable access to neurons with a known contribution to a behavior, I will adopt a rapidly-trained BMI- based learning task in which a mouse learns to activate a single target cortical neuron in a specific context. I will use high-bandwidth GETI imaging to study how the target neuron’s synaptic inputs and input-output operations change with learning. Moreover, I will adapt the BMI task to instead train neurons to perform an experimenter-selected input-output operation, to thereby investigate what types of input-output operations individual neurons can learn. These technologies combined will establish a new experimental paradigm with nearly limitless possibilities for studying neural computation and learning. I will use these tools to ask: 1) How are behaviorally- relevant input-output operations - the individual steps of neural algorithms - implemented within the cortex? 2) How do cortical neurons learn to perform a specific input-output operation? 3) What operations can individual cortical neurons learn to perform? and 4) Can we use the resulting knowledge to develop more effective BMIs?

Project Narrative I propose to develop technologies for measuring input-output operations of neurons and their circuits, a novel paradigm that would aid the study of all brain functions. I will apply these technologies to better understand how neurons in the cortex learn to interact with a brain-machine interface. This understanding will be particularly important to realize the therapeutic potential of brain machine interfaces in humans and assist patients with sensory or motor impairments.