Project Summary
To accommodate a vast amount of sensory information, the cortex selectively distills perceptually and behaviorally
salient features from stimuli. The visual system uses visual segmentation, the process of separating a visual scene
into individual elements, to achieve this. Mechanistically, visual segmentation is enabled by “surround suppres-
sion”, the attenuation of neural responses when a stimulus’ properties — such as orientation, contrast, or phase
— are homogeneous. For orientation, neurons are maximally suppressed when a stimulus is iso-oriented with its
surround and is relieved when they are orthogonal. Therefore, the orientation dependence of surround suppression
is an important cortical computation that underlies accurate and efficient visual processing of stimuli. However,
the exact mechanisms and organization of neural circuits responsible for generating oriented surround suppression
is unknown. This proposal will describe and test, in vivo, a detailed mechanistic circuit model explaining how
oriented surround suppression is generated in the cortex. The primary hypothesis is that orientation-dependent
surround suppression results from feedback connections from higher visual areas onto inhibitory interneurons in
the primary visual cortex. Specifically, I propose that neurons in the mouse higher visual area lateromedial (LM)
synapse onto somatostatin-expressing (SST) interneurons in V1 — a key mediator of surround suppression — such
that cells with similar orientation preferences are preferentially connected. In order to test this, I will exploit a
novel two-photon holographic mesoscope that can record and manipulate neurons across multiple brain regions
for the first time. With this microscope, I will holographically photostimulate excitatory neurons in LM, either
sequentially or simultaneously, while recording from multiple cell classes in V1. First, I will use this technique
to test the connectivity between single excitatory neurons in LM and their SST targets in V1 to determine how
orientation preference governs connectivity. Then, I will activate ensembles of iso-oriented LM neurons to generate
surround suppression when there should not be any. Together, these experiments will advance our fundamental
understanding of how neural architecture governs cortical computations in surround suppression. Consequently,
these results will also shed light on how neural circuits across brain areas coordinate to drive visual perception.
The proposed work not only improves our understanding of neural architecture, but also how disruptions in these
precise networks can lead to a number of psychiatric and neurological diseases.