Bay Area Vision Research Day (BAVRD) 2019

Cortical neurons pool information when signals are weak to optimize detection, whereas they differentiate input when signals are strong to optimize localization. The neural circuits that mediate this context-dependent switch from integration to differentiation are not known. Using two photon imaging and optogenetics in the primary visual cortex (V1) we identified a contrast-dependent disinhibitory circuit that dynamically gates how pyramidal neurons integrate over spatial context. Low contrast stimuli preferentially recruit VIP inhibitory interneurons, which in turn suppress SST interneurons, to prevent surround suppression for low contrast stimuli. Conversely, high contrast stimuli preferentially recruit SST neurons and suppress VIP neurons, enabling SST neurons to suppress the rest of the cortical network. Based on these findings, we propose that the VIP disinhibitory circuit dynamically matches the integrative mode of cortical networks to fast fluctuations in perceptual and computational demands to optimize sensory perception.

Glaucoma and other ocular injuries lead to significant retinal neuron loss, optic nerve damage, vision loss, and eventual blindness if not adequately managed. No current treatment can successfully regenerate lost retinal ganglion cell neurites and thereby repair damaged retina-brain connections. Past research focused on identifying and blocking extracellular factors that prevent neuronal regrowth, so little is known about neuronal intrinsic mechanisms that stimulate optic nerve regeneration. Recent technological innovations have enabled assessment of the differential regenerative abilities of retinal ganglion cell subtypes. In an optic nerve crush model, past work from this laboratory revealed that a subclass of RGCs, named α-RGCs, are selectively regenerated after PTEN elimination.

This work also showed that an α-RGC subclass-specific molecule, osteopontin, in conjunction with IGF1, is a biologically safe reagent that promotes regeneration, replacing the PTEN elimination paradigm. To expand the horizon of this innovative approach and discover new strategies for axon regeneration, the goal of this proposal is to further characterize neuronal subtype heterogeneity and identify cell-intrinsic reagents that promote comprehensive regeneration. A set of innovative genomic, genetic, viral, imaging and electrophysiological approaches will be used to decode neuronal subtype diversity, to characterize RGC subtype differences in survival and regeneration after injury, and to reveal new molecular targets for efficient repair of individual RGC subtypes. Identifying the molecular mechanisms of this protection and regeneration will open the potential for expanding it to more RGC subtypes. The ultimate goal of this work is to identify a set of cell-intrinsic cues and key signaling nodes that promote axon regeneration.

First, we used single-cell RNA-sequencing analysis in conjunction with mouse genetics experiments to classify α-RGCs into four subtypes in vivo. These new markers will enable comparison of individual α-RGC subtypes, identification of subtype-specific intrinsic factors, and tests of their regenerative roles in a given α-RGC single subtype. Second, we aim at understanding how osteopontin regulates α-RGC preferential survival and selective regeneration potential in vivo, specific receptors for osteopontin will be identified in α-RGCs, with a focus on integrin subunits. Loss-of-function experiments will be used to test the role of these receptors in modulating a-RGC survival and axon regeneration after axotomy. Third, we applied the same factors, including osteopontin, and the same collection of RGC subclasses to evaluate an experimental glaucoma model involving chronic intraocular pressure elevation. The proposed studies will establish a transformative platform to discover new biological drugs to promote axon regeneration across multiple RGC subtypes, with the ultimate goal of restoring vision through optic nerve regeneration. Such findings are expected to offer insights into the regeneration of diverse neuronal subtypes in the brain and spinal cord, enabling discovery of novel therapeutics for neuronal repair in various forms of ocular and neurological diseases.

The past few years have seen a startling and troubling rise in the fake news phenomena in which everyone from individuals to nation sponsored entities can produce and distribute misinformation. The implications of fake news range from a misinformed public to an existential threat to democracy, and horrific violence. At the same time, recent and rapid advances in machine learning are making it easier than ever to create sophisticated and compelling fake images. videos, and audio recordings, making the fake news phenomena even more powerful and dangerous. I will provide an overview of the creation of these so called deep fakes, their risks to our society and democracy, and I will describe emerging techniques for detecting these fakes.

Behavior requires coordination between independent sensory and motor systems. For example, picking a ripe apple could involve elbow flexion or extension depending on the apple’s location relative to the hand. However, the same photoreceptors perceive the apple regardless of hand position, and the same spinal motoneurons innervate the biceps regardless of apple location. The ability to flexibly pair sensory input with the appropriate movement arises at the sensorimotor interface, where changes in motor effector position remap stimulus representations to the neuronal ensemble whose movement fields now contain the stimuli. Remapping spans species, modalities, motor systems, and brain areas as a canonical sensorimotor principle, yet its circuit mechanisms are unknown. In particular, it is unknown whether motor effector position remaps activity to drive the appropriate movement by exciting neurons within the recruited ensemble or by inhibiting neurons outside this ensemble. I will describe a novel mouse behavior we have developed in order to address this problem and our identification of a corollary discharge pathway that mediates remapping through inhibitory sculpting.

Understanding the functions of a brain region requires knowing the neural representations of its myriad inputs, local neurons, and outputs. With genetically encoded activity indicators, we can now use in vivo imaging to monitor neural activity at synaptic resolution. In this talk, I will discuss how we use advanced high-resolution optical microscopy methods to investigate visual processing in the mouse primary visual cortex (V1), including the response properties of thalamic boutons and cortical neurons in layers 1–6 of awake mouse V1. Using adaptive optics that allows accurate measurement of bouton activity deep in cortex, we found that around half of the boutons in the main thalamorecipient L4 carried orientation-tuned information and that their orientation and direction biases were also dominant in the L4 neuron population, suggesting that these neurons may inherit their selectivity from tuned thalamic inputs. I will also discuss a new class of L6 neurons that subserve cross callosal communication in the mouse V1.

This talk will introduce the Oz Vision project, an early-stage project that seeks to build a new type of color display based on scanning a laser over the human retina, to create perceptions of new colors impossible to see in the real world, to treat color blindness, and to enable a person to perceive higher dimensional color, e.g. 4D with IR over RGB.

Computational imaging involves the joint design of imaging system hardware and software, optimizing across the entire pipeline from acquisition to reconstruction. Computers can replace bulky and expensive optics by solving computational inverse problems. This talk will describe new microscopes that use computational imaging to enable 3D, aberration and phase measurement using simple hardware and advanced image reconstruction algorithms that are based on large-scale nonlinear non-convex optimization.

Two distinct anatomical pathways carry visual signals from the retina to the mammalian visual cortex. The geniculate pathway passes through the dorsolateral geniculate nucleus of the thalamus to innervate primary visual cortex. The collicular pathway passes through the superior colliculus and reaches higher visual cortical areas via the pulvinar nucleus of the thalamus. Using the mouse as a model system I will show that a higher visual cortical area extract motion from the visual environment using exclusively information originating from the superior colliculus. Since the collicular pathway is considered phylogenetically older than the geniculate pathway, cortical visual areas depending on this route for motion processing may be considered remnants of an ancestral visual cortex.

Computer vision has advanced rapidly with deep learning, achieving super-human performance on a few recognition benchmarks. At the core of the state-of-the-art approaches for image classification, object detection, and semantic instance segmentation is sliding window classification, engineered for computational efficiency. Such piecemeal analysis of visual perception often has trouble getting details right and fails miserably with occlusion. Human vision, on the other hand, thrives on occlusion, excels at seeing wholes and parts, and can recognize objects with very little supervision. I will describe our models on approaching human-level perceptual organization with no annotations at all or with lesser and fewer annotations, delivering more accurate and generalizing performance beyond recognition in a closed world.

Natural language use emerges from the interactions between people and it grounds into our perceptual world. I will discuss how two-player real-time language games can provide a source of grounded language data for exploring psycholinguistic models of language meaning and pragmatics use. The same paradigms yield training data for modern neural language models. I will explore these ideas in the domains of language for color and shape.