Valentin Dragoi - US grants

[unreadable] DESCRIPTION (provided by applicant): In the visual system, the information transmitted from the retina is analyzed and transformed by the visual cortex at multiple stages to construct an internal representation of the environment. It has long been suggested that the visual cortex is a passive filter that creates a static, spatial, representation of a visual scene via a hierarchical processing of sensory inputs from the two eyes. According to this view, variations in neuronal responses to identical stimuli were believed to reflect stochastic fluctuations, or noise. However, this view largely ignores the fact that, even when the eyes are closed, the brain is far from being silent. Indeed, several lines of evidence indicate that fluctuations in population activity cause cortical networks to wander through various states of excitability. Although it is acknowledged that fluctuations in ongoing activity change the state of cortical networks involved in stimulus processing, little is known about whether and how cortical states interact with incoming stimuli to influence visual representations, and subsequently behavioral performance. This proposal will examine the relationship between fluctuations in the ongoing activity of neurons in primary visual cortex (V1) and orientation coding (Aim 1), the influence of ongoing activity on the ability of populations of neurons to reliably encode orientation signals (Aim 2), and the relationship between ongoing activity, stimulus-evoked response, and behavioral performance during orientation discrimination (Aim 3). We believe that our research on state-dependent information processing in visual cortex has the potential to advance our understanding of the neuronal mechanisms underlying visual perception and learning, and, at the same time, help develop chronically-implantable human cortical prostheses to assist visually impaired people. [unreadable] [unreadable] [unreadable]

Although we know a great deal about how individual neurons encode information, how networks of cells represent incoming stimuli in population activity has rarely been investigated experimentally. Indeed, although improved multi-electrode techniques and theoretical models are now available to help answer this important question, exactly how cortical networks encode information to influence the accuracy of behavioral responses remains unknown. We will employ state-of-the-art electrophysiological and behavioral techniques to record simultaneously the activity of multiple neurons in multiple visual cortical areas of behaving monkey in combination with computational models of network function to understand how neural circuits produce emergent properties. We plan to examine for the first time how populations of neurons in different cortical areas encode information to influence behavioral decisions. We will perform experiments of a high degree of difficulty to simultaneously record neuronal activity in multiple cortical areas along the same cortical processing stream (e.g., V1, V4, and IT; infero-temporal cortex, or IT, is the terminal processing station of the 'object' pathway) while monkeys will perform an image discrimination task. These experiments will offer us the unique opportunity to examine the coding of visual information at each stage of visual processing and the flow of information between different cortical areas, and thus investigate how the entire network engaged in image processing works to create and update visual representations that are relevant for behavior. The experiments will also allow us to examine for the first time the contribution of different types of cells (e.g., excitatory and inhibitory) to population coding and their impact on behavioral performance. We will examine for the first time the relationship between the properties of the population code in multiple cortical areas and behavioral decisions. We will perform experiments of a high degree of difficulty by simultaneously record neuronal activity in several cortical areas along the same cortical processing stream during an image discrimination task. These studies will offer us the unique opportunity to investigate how interactions between multiple neuronal networks influence behavioral decisions.

ABSTRACT I propose a new technology that will fundamentally change our understanding of brain function and the relationship with behavior. Examining complex functions, such as vision, hearing, memory, attention, decision making, or sleep, have been traditionally performed by studying the brain of nonhuman primates in a laboratory environment in which the head and body are restrained while synthetic stimuli are presented on a computer monitor. However, it has become increasingly understood that studying the brain in a restrained laboratory rig poses severe limits on our capacity to understand the function of brain circuits. Indeed, when responses of individual neurons are measured in naturalistic settings they tend to be different compared to when they are measured in the laboratory environment. In addition, it has always been unclear whether phenomena well described in laboratory conditions, such as adaptation, learning, or decision making, can be replicated when animals freely move in their natural environment. To overcome these limitations, I will construct a wireless system that will allow neuroscientists to study cortical dynamics and plasticity at the population level while nonhuman primates are moving freely in their natural environment. Studying brain function in an unconstrained environment will open new opportunities. Phenomena that were difficult or impossible to observe in an experimental rig, such as foraging, sleep, or social behavior will now become possible to study. Our technology will allow us to acquire vastly more brain data than has ever been gathered in live subjects, at a much higher rate than is possible today. The large quantity of contiguous neural data recorded by such a system will be of great interest to clinicians and theorists studying general properties of normal and dysfunctional neural networks. Our system will lead not only to medical insights into the mechanisms of brain disorders, but also to practical applications for neuronal prosthetic devices.

? DESCRIPTION (provided by applicant): Examining neural circuits crucially relies on the ability to activate or silence individual circuit components to subsequently assess their impact on other parts of the circuit and their influence on behavior. Recent refinements of viral tools for gene delivery have allowed optogenetic methods to target cells based on specific cell types, localization, and connectivity. The physiological dissection of targeted circuits has been extremely successful in the mouse brain, but remains of limited use in non-human primate brain. We plan to develop and test a new generation of viral tools that will allow us to both activate and suppress different cell types in non-human primate models. To accomplish our aims we have assembled an expert team with complementary expertise composed of a biochemist and photobiologist (John Spudich), a molecular neuroscientist (Roger Janz), and a systems and computational neuroscientist (Valentin Dragoi). Our approach builds upon recently discovered anion-conducting channelrhodopsins (ACRs), which perform with perfect anion selectivity, photosensitivity orders of magnitude greater than current optogenetic rhodopsins, and enable highly efficient neuron hyperpolarization. We believe that our ACR constructs will open a new chapter in targeted neuro-suppression. In addition, we will use new neuron-activating (depolarizing) cation-conducting channelrhodopsins (CCRs) that have ~3-fold greater unitary conductance, faster recovery from excitation, and higher sodium selectivity than the commonly used channelrhodopsin-2. We will construct viral vectors encoding ACR-CCR pairs and, using spectrally different ACRs, ACR-ACR pairs, enabling efficient wavelength-selected neuron activation or suppression in large populations. The effectiveness of these viral vectors will be tested in cultured and in situ mouse neurons and in the primary visual cortex (V1) of behaving monkeys. Developing these powerful tools will be invaluable for probing neural circuits in non-human primate models, finally allowing the interrogation of microcircuits underlying primate cognitive function.

? DESCRIPTION (provided by applicant): The brain evolved complex recurrent networks to enable flexible behavior in a dynamic and uncertain world, but its computational strategies and underlying mechanisms remain poorly understood. We propose to uncover the network basis of neural computations in foraging, an ethologically relevant behavioral task that involves sensory integration, spatial navigation, memory, and complex decision-making. We will use large-scale electrical recordings from six relevant interconnected areas (visual cortical area V4, Area 7A, Entorhinal Cortex, Hippocampus, Parahippocampal gyrus, and Prefrontal Cortex) of freely behaving macaques. To track the neural network computations used in these ethologically relevant, natural tasks, we will exploit recent advances in both statistical data analysis and theories of neural computation. First, to characterize behavior, we will model relationships between task-relevant sensory, motor, and internal variables using graphical modeling. Animal behavior will be modeled in the framework of Partially Observable Markov Decision Processes (POMDP) and these models will provide predictions about which variables the animals use and how they interact. Second, once we have modeled the behaviorally relevant variables, we will use modern data analysis techniques to identify these variables from the patterns of neuronal responses, extracting the low- dimensional, task-relevant signals from the high-dimensional population activity. The time series of these low- dimensional neural representations will be used to analyze the transformation and flow of signals between different brain areas, using such measures as Directed Information. Finally, we will compare these neural analyses to predictions from the normative models of the foraging task. We hypothesize that neural representations of sensory and internal variables will exhibit the same causal and temporal relationships manifested in the behavioral model. By combining - for the first time - normative modeling, selective dimensionality reduction of neural population signals, and quantification of directed information flow, we will be able to identify the transformations within and between key brain areas that enact neural computations on complex natural tasks. The team project aims to produce a transformative view of distributed neural population coding, unifying ethologically crucial computations across multiple neural systems.

? DESCRIPTION (provided by applicant): Examining neural circuits crucially relies on the ability to activate or silence individual circuit components to subsequently assess their impact on other parts of the circuit and their influence on behavior. Recent refinements of viral tools for gene delivery have allowed optogenetic methods to target cells based on specific cell types, localization, and connectivity. The physiological dissection of targeted circuits has been extremely successful in the mouse brain, but remains of limited use in non-human primate brain. We plan to develop and test a new generation of viral tools that will allow us to both activate and suppress different cell types in non-human primate models. To accomplish our aims we have assembled an expert team with complementary expertise composed of a biochemist and photobiologist (John Spudich), a molecular neuroscientist (Roger Janz), and a systems and computational neuroscientist (Valentin Dragoi). Our approach builds upon recently discovered anion-conducting channelrhodopsins (ACRs), which perform with perfect anion selectivity, photosensitivity orders of magnitude greater than current optogenetic rhodopsins, and enable highly efficient neuron hyperpolarization. We believe that our ACR constructs will open a new chapter in targeted neuro-suppression. In addition, we will use new neuron-activating (depolarizing) cation-conducting channelrhodopsins (CCRs) that have ~3-fold greater unitary conductance, faster recovery from excitation, and higher sodium selectivity than the commonly used channelrhodopsin-2. We will construct viral vectors encoding ACR-CCR pairs and, using spectrally different ACRs, ACR-ACR pairs, enabling efficient wavelength-selected neuron activation or suppression in large populations. The effectiveness of these viral vectors will be tested in cultured and in situ mouse neurons and in the primary visual cortex (V1) of behaving monkeys. Developing these powerful tools will be invaluable for probing neural circuits in non-human primate models, finally allowing the interrogation of microcircuits underlying primate cognitive function.

PROJECT SUMMARY/ABSTRACT Although sleep is generally believed to be essential for survival, the neuronal mechanisms by which it affects brain function are largely unknown. Does sleep constitute a passive state in which the brain is `quiet' or does it play an important role for network coding and behavior in subsequent tasks? Even though the functional significance of sleep is not well understood, the available data suggest that significant improvements in learning and memory are found even after brief naps. A critical issue for understanding sleep function is whether and how it impacts the accuracy of neuronal network computations to improve behavioral performance. We will address these issues for the first time by examining whether brief sleep (20 min of rest) influences visual perceptual performance and the coding of visual information across neuronal populations. We propose to use multiple-electrode recording simultaneously in two visual cortical areas (V1 and V4) of awake behaving monkey to examine the dynamics and coding in neuronal populations before, during, and after sleep, and their impact on perceptual performance. Aim 1 will examine how sensory experience changes the structure of network activity during rest by determining (i) how task exposure modifies the distribution of neuronal correlations across networks during rest, and (ii) if cells that are coactivated during stimulus exposure are more likely to be reactivated during subsequent rest. Aim 2 will examine whether brief sleep influences subsequent stimulus coding by individual neurons and networks by determining (i) whether single neuron discrimination performance is improved after rest, and (ii) whether and how correlated activity across the network is modified after rest. By decoding the population response we will determine (iii) whether neuronal populations encode more information after rest, and (iv) whether and how rest changes the synchrony between individual neurons and local population activity. Aim 3 will examine whether rest influences the relationship between neuronal and behavioral performance by determining (i) whether behavioral performance is improved after rest, (ii) whether the post-sleep neuronal and behavioral performance are correlated, (iii) whether LFP activity and spike-LFP synchronization during sleep are correlated with post-rest behavioral performance, (iv) whether there is a relationship between the amount of rest and the improvement in network and behavioral performance. Our research has the potential to advance our understanding of the neural mechanisms underlying rest and sleep and thus provide future solutions to ameliorate the detrimental effects of sleep disorder on cognitive performance, including practical applications for non-invasive neuronal prosthetic devices.

BIOSTATISTICS AND COMPUTATION MODULE ABSTRACT The Biostatistics and Computation Module (BCM) provides centralized resources and specialized services to members of the Vision Research Group to support all investigator projects. The module provides statistical consultation, mathematical, and computational support for NEI-funded investigators and their collaborators. In addition, the BCM will cooperate with the Imaging Module to evaluate new analytical tools and new software and hardware for image analysis. The module is staffed by an experienced biostatistician who works on a daily basis with faculty involved in vision research, and a computer specialist with extensive training in hardware and software design, image processing, computer graphics, and real-time data acquisition. The module will provide vision scientists and their collaborating investigators with expert support to design experiments and analyze data sets originating from anatomical, imaging, and electrophysiological experiments. Module staff will develop mathematical models describing novel phenomena arising from vision research projects, develop custom hardware and software, and provide support for devising and troubleshooting novel systems and interfaces designed to accelerate data acquisition and processing.

PROJECT SUMMARY/ABSTRACT Social interactions, a ubiquitous aspect of our everyday life, are critical to the health and survival of the species, but little is known about their underlying neural computations. The major limitation preventing our understanding of the neural underpinnings of social cognition is the lack of a suitable framework to allow us to study how it emerges in real time from interactions among brain networks. Indeed, examining the neural bases of complex social interactions has been traditionally performed by studying the brain of nonhuman primates in a laboratory environment in which the head and body are restrained while synthetic stimuli are presented on a computer monitor. However, it has become increasingly understood that studying the brain in spatially confined, artificial laboratory rigs poses severe limits on our capacity to understand the function of brain circuits. To overcome these limitations, we propose a novel approach to understand the neural underpinnings of social cognition. We will use a high-yield wireless system to study the cortical dynamics and plasticity of social interactions by recording population activity in multiple visual, temporal, and prefrontal cortical areas while nonhuman primates are interacting freely with their environment and with other animals. This new approach will enable us to uncover the dynamics of neuronal network activity that drives social interactions in an ethologically relevant behavioral task that involves sensory integration, memory, and complex decision-making. Our integrative project brings together innovative brain recording technologies and microelectronics together with large data sets analysis techniques. Our proposed research will constitute a paradigm shift by moving social neuroscience ? from simply observing animal behavior and recording the responses of single cells ? to a quantitative understanding of the distributed neuronal network encoding during social behavior in freely moving nonhuman primates performing rich naturalistic tasks. We anticipate that the large quantity of neural data recorded using our approach will be of great interest to clinicians and computational neuroscientists studying general properties of normal and dysfunctional neural networks, possibly leading to medical insights into the mechanisms of autism and attention deficit disorders that impair social interactions.

SUMMARY The brain transforms raw sensory input into perception and cognition, and this transformation relies on computations performed across neuronal circuits. Fortunately, the anatomy of cortical microcircuits in non- human primate models is much better understood today than decades ago. Indeed, sensory information travels in the cerebral cortex along feedforward and feedback pathways. While bottom-up feedforward connections have been extensively examined over the past several decades, the functional role of feedback projections continues to remain mysterious. Cortical feedback has been previously examined by reversibly inactivating higher cortex using a variety of methods, and measuring effects in single neurons in lower cortex. While important, previous studies modulating cortical feedback using techniques such as pharmacological inactivation, cortical cooling, electrical microstimulation, or transcranial magnetic stimulation, suffer from several key limitations, including poor spatial localization and temporal precision, and exclusive focus on single neuron responses. To overcome these limitations we will use optogenetic and multi-electrode electrophysiological methods in non-human primates to inactivate cortical feedback in real time based on cell localization, laminar location, and connectivity, and examine its impact of neural coding and behavior. To this end, we have chosen a major visual pathway involving primary visual cortex (V1) and mid-level visual cortex (V4). V4 neurons are believed to relay top-down signals related to behavioral context and attention to area V1 via direct feedback projections. Our working hypothesis is that feedback connections exhibit functional specificity and increase the population coding accuracy and communication among cortical neurons to improve behavioral performance. Our proposal will test the feasibility and validate the use of optogenetic methods in conjunction with multi-electrode recordings of population activity to directly address several of the desired capabilities for the next generation of neuroscience tools in non-human primates. We propose a new way to optogenetically control cortical feedback which will lay the groundwork for an interrogation of large-scale circuits at an unprecedented level of resolution. Thus, our proposal represents a significant step toward mapping the dynamic activity of relevant brain circuits in real time and understanding their impact on network coding and behavioral decisions.

PROJECT SUMMARY/ABSTRACT Stimuli presented too briefly to be noticed can nonetheless facilitate the perceptual processing of the same stimuli many minutes later. Whereas the phenomenon of subliminal priming has been known for decades, whether and how sensory information is encoded in the brain in the absence of awareness in a way that influences subsequent sensory processing across neuronal circuits remains a mystery. We will answer these questions for the first time by examining, at single cell resolution, whether exposure to subliminal stimuli influences perceptual performance and the coding of information across visual cortical populations. To accomplish this goal, we will use multiple-electrode recording of single-unit activity in macaque early and mid- level visual cortical areas (V1 and V4) and behavioral techniques to examine the dynamics and coding in neuronal populations during and after subliminal exposure, and their impact on perceptual performance. Aim 1 will investigate whether exposure to subliminal stimuli increases subsequent perceptual performance and the amount of information encoded in population activity. Our hypothesis is that subliminal exposure improves perceptual discrimination performance when stimuli are subsequently presented above the detectability threshold, and increases the amount of information extracted from the population response. Aim 2 will examine the mechanism of the improvement in neuronal and behavioral performance after subliminal exposure. We will first focus on causal experiments involving optogenetic inactivation which will test whether suppressing neuronal activity in visual cortex during the presentation of subliminal stimuli reduces the strength of subliminal priming. Cross-correlation analysis will subsequently test whether improved network and perceptual performance after subliminal exposure is consistent with a Hebbian mechanism underlying the increase in functional connectivity specifically for the neurons activated by subthreshold stimuli. Aim 3 will examine the impact of attention on the relationship between subliminal priming and neuronal and perceptual performance. We will test the novel hypothesis that spatial attention reduces the efficacy of subliminal priming ? exposure to unattended subliminal stimulation will be associated with a larger improvement in network coding and perceptual performance compared to exposure to attended information. In contrast, we expect that attention will increase the strength of supraliminal priming, i.e., exposure to attended subliminal stimulation will be associated with a larger improvement in network coding and perceptual performance compared to exposure to unattended information. Taken together, our proposed experiments can potentially advance our understanding of information coding in visual cortex by testing the limits of sensory experience and its relationship with perception, which will help develop effective therapies to treat the brain-based aspects of low vision conditions, such as amblyopia, age-related macular degeneration, diabetic retinopathy, or stroke.