Kevin Sit

Projects

My research takes a reverse-engineering approach to systems neuroscience—combining optical physiology, custom instrumentation, and computational modeling to ask how cortical circuits compute. A few of the threads I've worked on:

Bidirectional holographic optogenetics

In the Adesnik Lab, I designed, built, and tested a multicolor holographic two-photon microscope—the first of its kind—that can simultaneously excite and suppress hundreds of individually targeted neurons while imaging the same population. By independently tuning excitatory and inhibitory drive at single-cell resolution, the instrument makes it possible to test the necessity and sufficiency of specific neurons in cortical computation. I'm using it to show that the quenching of neural variability at stimulus onset arises from an intrinsic property of the cell, and to probe the surprising symmetry of patterned manipulations.

Visual contributions to spatial navigation

For my dissertation, I studied how visual landmarks anchor the brain's internal compass. Using two-photon imaging in retrosplenial cortex (RSC), I showed that visual input alone can drive the head-direction network in head-fixed mice, characterized the distinct feedforward signals arriving from thalamus and visual cortex, and identified a class of RSC neurons that integrates the two—leading to a circuit model for how vision keeps the internal compass registered to the world.

Functional organization of mouse visual cortex

My first graduate project asked how the mouse visual cortex is organized relative to that of primates. By adapting random dot kinematograms for mice, I mapped coherent-motion responses across higher visual areas and revealed both heterogeneity between areas and a striking visual-elevation gradient—evidence that the murine visual cortex is less modular and likely shaped by natural scene statistics.

Inhibitory circuits and contextual processing

Surround suppression—the attenuation of a neuron's response when a stimulus extends beyond its receptive field—is a fundamental cortical computation. Using two-photon holographic optogenetics to probe somatostatin (SST) interneurons, our work found that SST and excitatory neurons form a like-to-like, co-tuned connectivity pattern, and that manipulating these co-tuned cells shapes visual processing—clarifying how feature-specific inhibition sharpens cortical representations.