The Problem:

Sensorimotor transformation is a fundamental mechanism underlying the execution of most goal directed behaviors. Sensory inputs from peripheral organs, such as the skin, are relayed through neurons that synapse at various levels of the nervous system, including the spinal cord, brainstem, midbrain, thalamus, and ultimately, the cortex. Within cortical circuits, these sensory signals integrate with higher-order cognitive information to generate motor commands depending on the goal and environmental conditions. Neurons in the cortex perform these computations by communicating with other regions, both within the cortex and across the brain, through axonal projections. These communication pathways process specialized information depending on the regions they connect to and the required computations. This diversity in projection patterns can be used to classify distinct cortical cell types, which serve as fundamental building blocks of cortical circuits.

Despite advances in understanding these circuits, several critical questions remain unanswered. What unique information is processed by these cell types? How do they contribute to the encoding of specific behavioral features? What is the neural code, network dynamics, and computational mechanisms  that underlie their interactions? Addressing these is key to unraveling the principles governing sensorimotor control.

The Approach:

We aim to understand these principles by measuring, manipulating and modeling neural activity and circuit connectivity of projection-defined cortical neurons during naturalistic behaviors in awake, head-fixed mice. Mice are trained to perform naturalistic behavioral tasks requiring online sensorimotor control, such as hand-to-mouth feeding, oromanual manipulation, and reach-to-grasp tasks under head-fixed condition. We integrate state-of-the-art genetic, molecular and viral strategies with advanced techniques such as widefield and two-photon imaging, high-density electrophysiology and high-speed video recording to measure single-cell, population-level and inter-areal network dynamic of specific cell types in behaving mice. We use high resolution neuroanatomical tracing approaches to reveal circuit connectivity, while chemogenetic and optogenetic tools enable precise manipulation of neural activity. We apply Machine learning and AI techniques to analyze high dimensional neural and behavior data, and use computational modeling approaches to test hypotheses and interpret results. 

If you are interested in joining us on this journey or learning any of these techniques, reach out!