Research

Overview

We build technologies to record and control neural activity across the entire brain at millisecond timescales, the speed at which neurons actually operate. By combining whole-brain voltage imaging, single-cell perturbation, and biological experiments, we aim to uncover the principles by which vertebrate brains compute, learn, and adapt. We use larval zebrafish as a primary model system, which contains ~100,000 neurons in its transparent brain.

Voltage activity of ~20,000 neurons distributed across the entire zebrafish brain (top projection view), recorded using rsLSM 1.0, our first-generation imaging platform. Short video clip. (Note: the video may take a few moments to load)

Research directions

1. Whole-brain voltage imaging

Recording neural activity at the speed it actually happens. Many essential brain functions emerge from interactions across distributed neural populations and unfold at millisecond timescales. Understanding these processes requires measuring activity across the brain with both cellular resolution and temporal precision.

We develop imaging platforms that directly measure neuronal voltage activity across vertebrate brains at single-cell and millisecond resolution. Unlike calcium imaging, which reports slower and indirect signals, voltage imaging captures neural dynamics at the timescale of spikes and fast circuit computation.

Our first-generation remote-scanning light-sheet microscope (rsLSM 1.0) records voltage activity from more than 20,000 neurons across the larval zebrafish brain at single-cell resolution and ~200 Hz. We are now advancing next-generation systems toward broader coverage, improved signal quality, and deeper integration with biological experiments.

2. Whole-brain, single-cell resolution perturbation

Testing causality across the brain. Mapping neural dynamics alone does not reveal causality: neurons firing within milliseconds may be directly connected or may instead share common input, and distinct circuits can generate similar activity patterns. Establishing causal relationships requires temporally precise, cell-specific perturbations.

We are adapting holographic optogenetics for whole-brain perturbation at single-cell resolution and millisecond precision, and integrating it with whole-brain voltage imaging. This combined platform is designed to simultaneously read from and write to neurons across the entire vertebrate brain, enabling routine causal interrogation at whole-brain, single-cell, and single-spike level.

3. Neural mechanisms and modeling

From brain-wide activity and causality to mechanisms. We use these technologies to study the principles underlying brain function, neural computation, and learning. One process of particular interest is reinforcement learning, a fundamental mechanism by which animals adapt their behavior based on environmental feedback. To investigate this, we use a behavioral paradigm known as ROAST, in which larval zebrafish learn to execute directed turns through repeated reinforcement.

More broadly, we seek to connect brain-wide voltage dynamics with underlying circuit architecture (connectome). In collaboration with other teams, we aim to integrate large-scale functional activity maps with connectome structure, enabling more accurate models of neural circuits and, ultimately, more holistic digital representations of the vertebrate brain.

Google Sites

Report abuse