Understanding how patterns of electrical signals within the brain give rise to complex behaviour --- the 'neural code' --- is critical for not only discovering the neural basis of fundamental processes such as learning, decision making and cognition; but also for treating neuropsychiatric pathologies and motor impairments. The challenge is that unlike a computer, the neurons in the brain are highly recurrent, nonlinear and have long range interactions that vary across very short (millisecond) and very long (years) times-scales. The overall aim of this fellowship is to develop a tool that can dynamically interact with the brain at sufficient scale and resolution to reverse-engineer the neural circuitry.

Much like how self driving cars build a model of the world (e.g. other cars, road geometry, road surface) by dynamically interacting with it (accelerating, braking, imaging); the systems proposed here will dynamically interrogate populations of neurons --- at cellular-level resolution and with temporal dynamics comparable to individual action potentials --- to generate models of key neural processes and 'reverse engineer' the neural circuitry. Such a system would not only have radical implications for our understanding of basic processes such neural representation; but also could be used to 'drive' neuronal circuits, which would enable new frontiers in brain machine interfaces for neuropsychiatric prosthetics and clinical applications such as seizure prevention.

The gold-standard for interrogating electrical signals from individual neurons are electrophysiology techniques. However these require bulky electrical probes to be inserted into the brain, making it impossible to access large populations of neurons. Instead, I will address neuronal circuits using an entirely different modality: light. Nature has provided us with light sensitive proteins that can either optically report electrical changes with variations in fluorescence or generate an electrical signal under optical excitation. By engineering these proteins into neurons, it is possible to optically readout and control electrical activity in the brain.

This fellowship will perfect a new type of reporter called a 'voltage indicator' that directly reports action potentials rather than proxies for voltage change, such as calcium variations. Calcium changes happen two orders of magnitude slower than a typical action potential, meaning that critical timing information is lost. In contrast, voltage indicators observe action potentials in real-time, which is critical for dynamically interacting with the brain --- a self-driving car with a camera delay would quickly crash!

To keep pace with these rapid dynamics, I will develop ultra-fast optical hardware to readout and control electrical signals all within the time it takes an action potential to propagate, and at sufficient scale to access neuronal circuits. Moreover, by closing the loop between readout and control, it will be possible to trigger a neuron to spike, then 'track' the neurons that respond, thus determining the wiring diagram. This 'all-optical' approach enables high-throughput cellular resolution connectomics (i.e. functional connection mapping of circuits) in vivo, and would be transformative to our understanding of the structure of the nervous system, e.g. for identifying circuit defects in neuropsychiatric disorders.

This fellowship, hosted at the Wolfson Institute for Biomedical Research at UCL, builds tools to ask entirely new questions about the function and structure of the brain, which are not possible using existing technology. It would have wide ranging applications in neuroscience and beyond (including cardiac, renal, and hepatic physiology). The proposed fellowship is therefore very well aligned with the BBSRC priority area 'biosciences for health' under the 'technology development for the biosciences' responsive mode priority.

Technical Abstract:
The central aim of this fellowship is to develop a closed-loop platform for scalable electrophysiology in the optical domain: the 'All-Optical VoltageScope' (AOVS). The AOVS is designed for (1) highly sensitive optical readout of neuronal activity at cellular resolution via genetically encoded voltage indicators (GEVIs); (2) ultra-fast imaging capable of capturing action potentials at sub-millisecond timescales across neuronal microcircuits (~500 neurons); (3) cellular-level optogenetic control of large numbers of neurons; (4) 'trial-by-trial' closed-loop operation, enabling the activity of a single action potential to control the photoexcitation of large neuronal ensembles in real-time.

To achieve these ambitious goals, the first stage of the fellowship develops a novel high-speed imaging system, MUX-RAMP, that combines inertia free random access imaging for rapid beam steering, multiplexed excitation for increased bandwidth and temporal focusing for maximal signal extraction. Next, I apply the MUX-RAMP to image GEVIs expressed in the cortex of behaving mice. Finally, I combine imaging with optogenetic stimulation in a closed loop manner, leveraging ultra-fast electronics and high-speed light patterning.

This fellowship leverages an inherently interdisciplinary approach, borrowing from applied physics and electrical engineering but with a strong focus on applications in circuit neuroscience. This unique approach enables entirely new questions to be probed, such as the role that inhibition, subthreshold spiking and spike timing play in circuit computation. This is not possible using existing technologies and the AOVS therefore fills a major gap in the biosciences landscape. This fellowship therefore fits very well with the BBSRC priority area 'biosciences for health' under the 'technology development for the biosciences' responsive mode priority.